Github messages for voidlinux
 help / color / mirror / Atom feed
* [ISSUE] qemu: datadir is not set, uefi support is broken
@ 2021-01-16 10:40 gspe
  2021-01-19 21:22 ` ericonr
                   ` (64 more replies)
  0 siblings, 65 replies; 66+ messages in thread
From: gspe @ 2021-01-16 10:40 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1228 bytes --]

New issue by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965

Description:
this pr https://github.com/void-linux/void-packages/pull/24884 remove `--datadir=/usr/lib` but not set `--datadir=/usr/share` so now qemu package has broken support for uefi, for example this is the content of `60-edk2-x86_64.json` :
```
cat /usr/share/qemu/firmware/60-edk2-x86_64.json
{
    "description": "UEFI firmware for x86_64",
    "interface-types": [
        "uefi"
    ],
    "mapping": {
        "device": "flash",
        "executable": {
            "filename": "share/qemu/edk2-x86_64-code.fd",
            "format": "raw"
        },
        "nvram-template": {
            "filename": "share/qemu/edk2-i386-vars.fd",
            "format": "raw"
        }
    },
    "targets": [
        {
            "architecture": "x86_64",
            "machines": [
                "pc-i440fx-*",
                "pc-q35-*"
            ]
        }
    ],
    "features": [
        "acpi-s3",
        "amd-sev",
        "verbose-dynamic"
    ],
    "tags": [

    ]
}
```

as you can see the filename path is not correct ` "filename": "share/qemu/edk2-i386-vars.fd",`


^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
@ 2021-01-19 21:22 ` ericonr
  2021-01-19 22:51 ` gspe
                   ` (63 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ericonr @ 2021-01-19 21:22 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 192 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-763149824

Comment:
@gspe do you think you can make a PR to fix this?

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
  2021-01-19 21:22 ` ericonr
@ 2021-01-19 22:51 ` gspe
  2021-01-19 22:55 ` ericonr
                   ` (62 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: gspe @ 2021-01-19 22:51 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 164 bytes --]

New comment by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-763193945

Comment:
I'll make a PR tomorrow.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
  2021-01-19 21:22 ` ericonr
  2021-01-19 22:51 ` gspe
@ 2021-01-19 22:55 ` ericonr
  2021-01-20 17:55 ` gspe
                   ` (61 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ericonr @ 2021-01-19 22:55 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 431 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-763195669

Comment:
For the record, qemu has switched to meson but hid that by including a copy of meson in their source tree and calling that from their 2k line configure script (their `meson.build` also has 2k lines, yay).

If you want a challenge, you can try to move qemu over to the meson build style.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (2 preceding siblings ...)
  2021-01-19 22:55 ` ericonr
@ 2021-01-20 17:55 ` gspe
  2021-01-20 17:55 ` gspe
                   ` (60 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: gspe @ 2021-01-20 17:55 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 799 bytes --]

New comment by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-763825068

Comment:
@ericonr I've tried to make a PR for qemu but I have a problem: building the package fails tests.
I have updated my void-packages repo and tried to build the original qemu in void-packages without any changes and it also fails test.

Regarding the meson build style I have read some documentation and according with the changelog https://wiki.qemu.org/ChangeLog/5.2#Build_Information the switch to meson is not complete at the moment:

> The build system is now partly based on Meson. However, building is still done with configure and make as in previous versions of QEMU.
indeed some building options are still not present in `meson_options.txt` file.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (3 preceding siblings ...)
  2021-01-20 17:55 ` gspe
@ 2021-01-20 17:55 ` gspe
  2021-01-21  1:41 ` ericonr
                   ` (59 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: gspe @ 2021-01-20 17:55 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 801 bytes --]

New comment by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-763825068

Comment:
@ericonr I've tried to make a PR for qemu but I have a problem: building the package fails tests.
I have updated my void-packages repo and tried to build the original qemu in void-packages without any changes and it also fails test.

Regarding the meson build style I have read some documentation and according with the changelog https://wiki.qemu.org/ChangeLog/5.2#Build_Information the switch to meson is not complete at the moment:

> The build system is now partly based on Meson. However, building is still done with configure and make as in previous versions of QEMU.

indeed some building options are still not present in `meson_options.txt` file.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (4 preceding siblings ...)
  2021-01-20 17:55 ` gspe
@ 2021-01-21  1:41 ` ericonr
  2021-01-21  1:42 ` ericonr
                   ` (58 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ericonr @ 2021-01-21  1:41 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 293 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764168463

Comment:
> building the package fails tests.

I doubt some can even be run on a container, so it should be fine to gate them with `"$XBPS_CHECK_PKGS" = full`

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (5 preceding siblings ...)
  2021-01-21  1:41 ` ericonr
@ 2021-01-21  1:42 ` ericonr
  2021-01-21 21:02 ` gspe
                   ` (57 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ericonr @ 2021-01-21  1:42 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 341 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764168463

Comment:
> building the package fails tests.

I doubt some can even be run on a container, so it should be fine to gate them with `"$XBPS_CHECK_PKGS" = full`

And ok, we can stick with configure for now.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (6 preceding siblings ...)
  2021-01-21  1:42 ` ericonr
@ 2021-01-21 21:02 ` gspe
  2021-01-21 21:10 ` ericonr
                   ` (56 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: gspe @ 2021-01-21 21:02 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 7313 bytes --]

New comment by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764939792

Comment:
I've used `"$XBPS_CHECK_PKGS" = full` option but it doesn't make any difference I still got errors during checks
this is pastebin of them https://pastebin.com/VUCnbbTY

This is just one of them in case of pastebin doesn't works
```
TEST    iotest-qcow2: 229 [fail]
QEMU          -- "/builddir/qemu-5.2.0/build/tests/qemu-iotests/../../qemu-system-x86_64" -nodefaults -display none -accel qtest
QEMU_IMG      -- "/builddir/qemu-5.2.0/build/tests/qemu-iotests/../../qemu-img"
QEMU_IO       -- "/builddir/qemu-5.2.0/build/tests/qemu-iotests/../../qemu-io"  --cache writeback --aio threads -f qcow2
QEMU_NBD      -- "/builddir/qemu-5.2.0/build/tests/qemu-iotests/../../qemu-nbd"
IMGFMT        -- qcow2 (compat=1.1)
IMGPROTO      -- file
PLATFORM      -- Linux/x86_64  5.9.16_1
TEST_DIR      -- /builddir/qemu-5.2.0/build/tests/qemu-iotests/scratch
SOCK_DIR      -- /tmp/tmp.gRja3PqgOn
SOCKET_SCM_HELPER -- /builddir/qemu-5.2.0/build/tests/qemu-iotests/socket_scm_helper

--- /builddir/qemu-5.2.0/tests/qemu-iotests/229.out	2020-12-08 17:59:44.000000000 +0100
+++ /builddir/qemu-5.2.0/build/tests/qemu-iotests/229.out.bad	2021-01-21 10:50:21.549119104 +0100
@@ -8,7 +8,14 @@

 === Starting drive-mirror, causing error & stop  ===

-{'execute': 'drive-mirror', 'arguments': {'device': 'testdisk', 'format': 'IMGFMT', 'target': 'blkdebug:TEST_DIR/blkdebug.conf:TEST_DIR/d.IMGFMT', 'sync': 'full', 'mode': 'existing', 'on-source-error': 'stop', 'on-target-error': 'stop' }}
+{'execute': 'drive-mirror',
+                 'arguments': {'device': 'testdisk',
+                               'format': 'IMGFMT',
+                               'target': 'blkdebug:TEST_DIR/blkdebug.conf:TEST_DIR/d.IMGFMT',
+                               'sync':   'full',
+                               'mode':   'existing',
+                               'on-source-error': 'stop',
+                               'on-target-error': 'stop' }}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "created", "id": "testdisk"}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "running", "id": "testdisk"}}
 {"return": {}}
@@ -17,7 +24,9 @@

 === Force cancel job paused in error state  ===

-{'execute': 'block-job-cancel', 'arguments': { 'device': 'testdisk', 'force': true}}
+{'execute': 'block-job-cancel',
+                 'arguments': { 'device': 'testdisk',
+                                'force': true}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "running", "id": "testdisk"}}
 {"return": {}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "aborting", "id": "testdisk"}}
  TEST    iotest-qcow2: 244
  TEST    iotest-qcow2: 249 [fail]
QEMU          -- "/builddir/qemu-5.2.0/build/tests/qemu-iotests/../../qemu-system-x86_64" -nodefaults -display none -accel qtest
QEMU_IMG      -- "/builddir/qemu-5.2.0/build/tests/qemu-iotests/../../qemu-img"
QEMU_IO       -- "/builddir/qemu-5.2.0/build/tests/qemu-iotests/../../qemu-io"  --cache writeback --aio threads -f qcow2
QEMU_NBD      -- "/builddir/qemu-5.2.0/build/tests/qemu-iotests/../../qemu-nbd"
IMGFMT        -- qcow2 (compat=1.1)
IMGPROTO      -- file
PLATFORM      -- Linux/x86_64  5.9.16_1
TEST_DIR      -- /builddir/qemu-5.2.0/build/tests/qemu-iotests/scratch
SOCK_DIR      -- /tmp/tmp.gRja3PqgOn
SOCKET_SCM_HELPER -- /builddir/qemu-5.2.0/build/tests/qemu-iotests/socket_scm_helper

--- /builddir/qemu-5.2.0/tests/qemu-iotests/249.out	2020-12-08 17:59:44.000000000 +0100
+++ /builddir/qemu-5.2.0/build/tests/qemu-iotests/249.out.bad	2021-01-21 10:50:22.114119114 +0100
@@ -7,24 +7,29 @@

 === Send a write command to a drive opened in read-only mode (1)

-{ 'execute': 'human-monitor-command', 'arguments': {'command-line': 'qemu-io none0 "aio_write 0 2k"'}}
+{ 'execute': 'human-monitor-command',
+       'arguments': {'command-line': 'qemu-io none0 "aio_write 0 2k"'}}
 {"return": "Block node is read-onlyrn"}

 === Run block-commit on base using an invalid filter node name

-{ 'execute': 'block-commit', 'arguments': {'job-id': 'job0', 'device': 'none1', 'top-node': 'int', 'filter-node-name': '1234'}}
+{ 'execute': 'block-commit',
+       'arguments': {'job-id': 'job0', 'device': 'none1', 'top-node': 'int',
+                     'filter-node-name': '1234'}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "created", "id": "job0"}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "null", "id": "job0"}}
 {"error": {"class": "GenericError", "desc": "Invalid node name"}}

 === Send a write command to a drive opened in read-only mode (2)

-{ 'execute': 'human-monitor-command', 'arguments': {'command-line': 'qemu-io none0 "aio_write 0 2k"'}}
+{ 'execute': 'human-monitor-command',
+       'arguments': {'command-line': 'qemu-io none0 "aio_write 0 2k"'}}
 {"return": "Block node is read-onlyrn"}

 === Run block-commit on base using the default filter node name

-{ 'execute': 'block-commit', 'arguments': {'job-id': 'job0', 'device': 'none1', 'top-node': 'int'}}
+{ 'execute': 'block-commit',
+       'arguments': {'job-id': 'job0', 'device': 'none1', 'top-node': 'int'}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "created", "id": "job0"}}
 {"timestamp": {"seconds":  TIMESTAMP, "microseconds":  TIMESTAMP}, "event": "JOB_STATUS_CHANGE", "data": {"status": "running", "id": "job0"}}
 {"return": {}}
@@ -36,6 +41,7 @@

 === Send a write command to a drive opened in read-only mode (3)

-{ 'execute': 'human-monitor-command', 'arguments': {'command-line': 'qemu-io none0 "aio_write 0 2k"'}}
+{ 'execute': 'human-monitor-command',
+       'arguments': {'command-line': 'qemu-io none0 "aio_write 0 2k"'}}
 {"return": "Block node is read-onlyrn"}
 *** done
  TEST    iotest-qcow2: 251
  TEST    iotest-qcow2: 252
  TEST    iotest-qcow2: 256
  TEST    iotest-qcow2: 259 [not run]
  TEST    iotest-qcow2: 265
  TEST    iotest-qcow2: 267
  TEST    iotest-qcow2: 268
  TEST    iotest-qcow2: 271
  TEST    iotest-qcow2: 283
  TEST    iotest-qcow2: 287
  TEST    iotest-qcow2: 290
  TEST    iotest-qcow2: 292
  TEST    iotest-qcow2: 299
  TEST    iotest-qcow2: 309
Not run: 259
Failures: 117 127 140 141 143 156 161 191 229 249
Failed 10 of 122 iotests
make[1]: *** [/builddir/qemu-5.2.0/tests/Makefile.include:144: check-block] Error 1
make[1]: Leaving directory '/builddir/qemu-5.2.0/build'
make: *** [GNUmakefile:11: check] Error 2
=> ERROR: qemu-5.2.0_1: do_check: '${make_cmd} ${make_check_args} ${make_check_target}' exited with 2
=> ERROR:   in do_check() at common/build-style/configure.sh:32
```

Maybe depends on my hardware, I'm using a ryzen 7 or maybe it's better to wait that the c/c++ upgrade is finished.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (7 preceding siblings ...)
  2021-01-21 21:02 ` gspe
@ 2021-01-21 21:10 ` ericonr
  2021-01-21 21:15 ` gspe
                   ` (55 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ericonr @ 2021-01-21 21:10 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 215 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764943567

Comment:
How are you running the build?

Also what's the template looking like?

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (8 preceding siblings ...)
  2021-01-21 21:10 ` ericonr
@ 2021-01-21 21:15 ` gspe
  2021-01-21 21:19 ` ericonr
                   ` (54 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: gspe @ 2021-01-21 21:15 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 312 bytes --]

New comment by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764946414

Comment:
I'm running the standart command to build packages `./xbps-src -f pkg qemu`

The template is the original one in srcpkgsc, I haven't made any changes to it for this test.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (9 preceding siblings ...)
  2021-01-21 21:15 ` gspe
@ 2021-01-21 21:19 ` ericonr
  2021-01-21 21:21 ` gspe
                   ` (53 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ericonr @ 2021-01-21 21:19 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 202 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764948290

Comment:
Do you have anything about `XBPS_CHECK_PKGS` in `etc/conf`?

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (10 preceding siblings ...)
  2021-01-21 21:19 ` ericonr
@ 2021-01-21 21:21 ` gspe
  2021-01-21 21:22 ` gspe
                   ` (52 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: gspe @ 2021-01-21 21:21 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 729 bytes --]

New comment by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764949476

Comment:
This is my etc/conf
```
cat etc/conf
# [OPTIONAL]
# Enable building package locally that are restricted legally for redistribution.
# NOTE: you can't distribute the sources or binaries for such kind of packages.
#
XBPS_ALLOW_RESTRICTED=yes

# [OPTIONAL]
# Number of parallel jobs to execute when building packages that
# use make(1) or alike commands.
#
XBPS_MAKEJOBS=16

# [OPTIONAL]
# Enable running the (optional) do_check() function of a package.
# When set to 'full', will enable further testing for some packages.
#
# XBPS_CHECK_PKGS=yes
XBPS_CHECK_PKGS=full
```

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (11 preceding siblings ...)
  2021-01-21 21:21 ` gspe
@ 2021-01-21 21:22 ` gspe
  2021-01-21 21:23 ` gspe
                   ` (51 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: gspe @ 2021-01-21 21:22 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 731 bytes --]

New comment by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764949476

Comment:
This is my `etc/conf`
```
cat etc/conf
# [OPTIONAL]
# Enable building package locally that are restricted legally for redistribution.
# NOTE: you can't distribute the sources or binaries for such kind of packages.
#
XBPS_ALLOW_RESTRICTED=yes

# [OPTIONAL]
# Number of parallel jobs to execute when building packages that
# use make(1) or alike commands.
#
XBPS_MAKEJOBS=16

# [OPTIONAL]
# Enable running the (optional) do_check() function of a package.
# When set to 'full', will enable further testing for some packages.
#
# XBPS_CHECK_PKGS=yes
XBPS_CHECK_PKGS=full
```

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (12 preceding siblings ...)
  2021-01-21 21:22 ` gspe
@ 2021-01-21 21:23 ` gspe
  2021-01-21 21:24 ` gspe
                   ` (50 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: gspe @ 2021-01-21 21:23 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 312 bytes --]

New comment by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764946414

Comment:
I'm running the standard command to build packages `./xbps-src -f pkg qemu`

The template is the original one in srcpkgsc, I haven't made any changes to it for this test.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (13 preceding siblings ...)
  2021-01-21 21:23 ` gspe
@ 2021-01-21 21:24 ` gspe
  2021-01-21 21:28 ` ericonr
                   ` (49 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: gspe @ 2021-01-21 21:24 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 311 bytes --]

New comment by gspe on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764946414

Comment:
I'm running the standard command to build packages `./xbps-src -f pkg qemu`

The template is the original one in srcpkgs, I haven't made any changes to it for this test.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (14 preceding siblings ...)
  2021-01-21 21:24 ` gspe
@ 2021-01-21 21:28 ` ericonr
  2021-01-24 15:14 ` ackalker
                   ` (48 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ericonr @ 2021-01-21 21:28 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 314 bytes --]

New comment by ericonr on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-764952498

Comment:
> XBPS_CHECK_PKGS=full

Well there you go :P 

It's always running full tests. You can run them selectively with `-K`, that way you don't need that flag in `etc/conf`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (15 preceding siblings ...)
  2021-01-21 21:28 ` ericonr
@ 2021-01-24 15:14 ` ackalker
  2021-01-24 15:17 ` ackalker
                   ` (47 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 15:14 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1182 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766365866

Comment:
Back on topic for this issue: I've had a look at the Qemu source, in particular the `meson.build` file I suspect is responsible for the problem: 
https://github.com/qemu/qemu/blob/553032db17440f8de011390e5a1cfddd13751b0b/pc-bios/descriptors/meson.build#L11

I don't see any particular problem with it, but I'm in no way a Meson expert. AFAICT the `configure_file` is supposed to take a file, change any references to `@DATADIR@` in it to the contents of `qemu_datadir`, and install it into the directory named by the contatenation of the value of `qemu_datadir`, a `/`, and the string `firmware`, which seems mostly what is happening during the build, except for some reason `@DATADIR@` gets substituted with `share/qemu` and not `/usr/share/qemu`, i.e. it is missing the prefix: `/usr`.
I find this very strange, as in other places, `qemu_datadir` does seem to evaluate to `/usr/share/qemu`.
Maybe some interaction going on here? Is there perhaps an environment variable `DATADIR` whose value is overriding that of the substitution?

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (16 preceding siblings ...)
  2021-01-24 15:14 ` ackalker
@ 2021-01-24 15:17 ` ackalker
  2021-01-24 15:19 ` ackalker
                   ` (46 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 15:17 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1219 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766365866

Comment:
Back on topic for this issue: I've had a look at the Qemu source, in particular the `meson.build` file I suspect is responsible for the problem: [pc-bios/descriptors/meson.build#L11](https://github.com/qemu/qemu/blob/553032db17440f8de011390e5a1cfddd13751b0b/pc-bios/descriptors/meson.build#L11)

I don't see any particular problem with it, but I'm in no way a Meson expert. AFAICT the `configure_file` is supposed to take a file, change any references to `@DATADIR@` in it to the contents of `qemu_datadir`, and install it into the directory named by the contatenation of the value of `qemu_datadir`, a `/`, and the string `firmware`, which seems mostly what is happening during the build, except for some reason `@DATADIR@` gets substituted with `share/qemu` and not `/usr/share/qemu`, i.e. it is missing the prefix: `/usr`.
I find this very strange, as in other places, `qemu_datadir` does seem to evaluate to `/usr/share/qemu`.
Maybe some interaction going on here? Is there perhaps an environment variable `DATADIR` whose value is overriding that of the substitution?

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (17 preceding siblings ...)
  2021-01-24 15:17 ` ackalker
@ 2021-01-24 15:19 ` ackalker
  2021-01-24 15:21 ` ackalker
                   ` (45 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 15:19 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1236 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766365866

Comment:
Back on topic for this issue: I've had a look at the Qemu source, in particular the `meson.build` file I suspect is responsible for the problem: [pc-bios/descriptors/meson.build#L11](https://github.com/qemu/qemu/blob/553032db17440f8de011390e5a1cfddd13751b0b/pc-bios/descriptors/meson.build#L11)

I don't see any particular problem with it, but I'm in no way a Meson expert.

AFAICT the `configure_file` action is supposed to take a file, change any references to `@DATADIR@` in its contents to the value of `qemu_datadir`, and install it into the directory named by the contatenation of the value of `qemu_datadir`, a `/`, and the string `firmware`, which seems mostly what is happening during the build, except for some reason `@DATADIR@` gets substituted with `share/qemu` and not `/usr/share/qemu`, i.e. it is missing the prefix: `/usr`.
I find this very strange, as in other places, `qemu_datadir` does seem to evaluate to `/usr/share/qemu`.
Maybe some interaction going on here? Is there perhaps an environment variable `DATADIR` whose value is overriding that of the substitution?

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (18 preceding siblings ...)
  2021-01-24 15:19 ` ackalker
@ 2021-01-24 15:21 ` ackalker
  2021-01-24 15:28 ` ackalker
                   ` (44 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 15:21 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1319 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766365866

Comment:
Back on topic for this issue: I've had a look at the Qemu source, in particular the `meson.build` file I suspect is responsible for the problem: [pc-bios/descriptors/meson.build#L11](https://github.com/qemu/qemu/blob/553032db17440f8de011390e5a1cfddd13751b0b/pc-bios/descriptors/meson.build#L11)

I don't see any particular problem with it, but I'm in no way a Meson expert.

AFAICT the `configure_file` action is supposed to take a file, change any references to `@DATADIR@` in its contents to the value of `qemu_datadir`, and install it into the directory named by the contatenation of the value of `qemu_datadir`, a `/`, and the string `firmware`, which seems mostly what is happening during the build. Files get installed into `/usr/share/qemu/firmware` (ignoring `$DESTDIR` for now), except for some reason `@DATADIR@` gets substituted with `share/qemu` and not `/usr/share/qemu`, i.e. it is missing the prefix: `/usr`.
I find this very strange, as in other places, `qemu_datadir` does seem to evaluate to `/usr/share/qemu`.
Maybe some interaction going on here? Is there perhaps an environment variable `DATADIR` whose value is overriding that of the substitution?

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (19 preceding siblings ...)
  2021-01-24 15:21 ` ackalker
@ 2021-01-24 15:28 ` ackalker
  2021-01-24 15:42 ` ackalker
                   ` (43 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 15:28 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1332 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766365866

Comment:
Back on topic for this issue: I've had a look at the Qemu source, in particular the `meson.build` file I suspect is responsible for the problem: [pc-bios/descriptors/meson.build#L11](https://github.com/qemu/qemu/blob/553032db17440f8de011390e5a1cfddd13751b0b/pc-bios/descriptors/meson.build#L11)

I don't see any particular problem with it, but I'm in no way a Meson expert.

AFAICT the `configure_file` action is supposed to take a file, change any references to `@DATADIR@` in its contents to the value of `qemu_datadir`, and install it into the directory named by the contatenation of the value of `qemu_datadir`, a `/`, and the string `firmware`, which seems mostly what is happening during the build. Files get installed into `/usr/share/qemu/firmware` (ignoring `$DESTDIR` for now), except for some reason `@DATADIR@` gets substituted with `share/qemu` and not `/usr/share/qemu`, i.e. it is missing the prefix: `/usr`.
I find this very strange, as in other places, `qemu_datadir` does seem to evaluate to `/usr/share/qemu`.
Is there maybe some bad interaction going on here? Is there perhaps an environment variable `DATADIR` whose value is overriding that of the substitution?

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (20 preceding siblings ...)
  2021-01-24 15:28 ` ackalker
@ 2021-01-24 15:42 ` ackalker
  2021-01-24 15:46 ` ackalker
                   ` (42 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 15:42 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 322 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
This commit seems to be the place where `qemu_datadir` gets introduced: qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29. The commit message is puzzling to me to say the least.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (21 preceding siblings ...)
  2021-01-24 15:42 ` ackalker
@ 2021-01-24 15:46 ` ackalker
  2021-01-24 15:48 ` ackalker
                   ` (41 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 15:46 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 388 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
The following 2 commits seems to be the place where `qemu_datadir` gets introduced: qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29, qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580. The commit messages are puzzling to me to say the least.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (22 preceding siblings ...)
  2021-01-24 15:46 ` ackalker
@ 2021-01-24 15:48 ` ackalker
  2021-01-24 17:39 ` ackalker
                   ` (40 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 15:48 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 389 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
The following 2 commits seems to be the place where `qemu_datadir` gets introduced: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.
The commit messages are puzzling to me to say the least.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (23 preceding siblings ...)
  2021-01-24 15:48 ` ackalker
@ 2021-01-24 17:39 ` ackalker
  2021-01-24 17:39 ` ackalker
                   ` (39 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:39 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 637 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: `configure_file()` expects a path _relative to `prefix` for its `install` keyword argument, but the substitution inside the JSON files needs to be done with an _absolute_ path. This can never be accomplished by using the same value (of `qemu_datadir`) for both.
The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (24 preceding siblings ...)
  2021-01-24 17:39 ` ackalker
@ 2021-01-24 17:39 ` ackalker
  2021-01-24 17:45 ` ackalker
                   ` (38 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:39 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 651 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: `configure_file()` expects a path _relative to the value of `prefix`_ for its `install` keyword argument, but the substitution inside the JSON files needs to be done with an _absolute_ path. This can never be accomplished by using the same value (of `qemu_datadir`) for both.
The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (25 preceding siblings ...)
  2021-01-24 17:39 ` ackalker
@ 2021-01-24 17:45 ` ackalker
  2021-01-24 17:46 ` ackalker
                   ` (37 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:45 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 830 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative to the value of `prefix`, but the `configuration` keyword argument which manages the substitution inside the JSON files needs to be given an _absolute_ path for it to work. This can never be accomplished by using the same value (of `qemu_datadir`) for both.
The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (26 preceding siblings ...)
  2021-01-24 17:45 ` ackalker
@ 2021-01-24 17:46 ` ackalker
  2021-01-24 17:47 ` ackalker
                   ` (36 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:46 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 831 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which manages the substitution inside the JSON files needs to be given an _absolute_ path for it to work. This can never be accomplished by using the same value (of `qemu_datadir`) for both.
The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (27 preceding siblings ...)
  2021-01-24 17:46 ` ackalker
@ 2021-01-24 17:47 ` ackalker
  2021-01-24 17:48 ` ackalker
                   ` (35 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:47 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 847 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which supplies the rules for the substitutions inside the JSON files needs to be given an _absolute_ path for it to work. This can never be accomplished by using the same value (of `qemu_datadir`) for both.
The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (28 preceding siblings ...)
  2021-01-24 17:47 ` ackalker
@ 2021-01-24 17:48 ` ackalker
  2021-01-24 17:51 ` ackalker
                   ` (34 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:48 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 868 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which supplies the rules for the substitutions inside the JSON files needs to be given an _absolute_ path for the substitutions to be correct. This can never be accomplished by using the same value (of `qemu_datadir`) for both.
The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (29 preceding siblings ...)
  2021-01-24 17:48 ` ackalker
@ 2021-01-24 17:51 ` ackalker
  2021-01-24 17:52 ` ackalker
                   ` (33 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:51 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 887 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which supplies the rules for the substitutions inside the JSON files needs to be given an _absolute_ path for the substitutions to be correct. Because `qemu_datadir` is relative, bo `configuration` and `install_dir` end up using relative paths.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (30 preceding siblings ...)
  2021-01-24 17:51 ` ackalker
@ 2021-01-24 17:52 ` ackalker
  2021-01-24 17:52 ` ackalker
                   ` (32 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:52 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 897 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which supplies the rules for the substitutions inside the JSON files needs to be given an _absolute_ path for the substitutions to be correct. Because `qemu_datadir` is relative, both `configuration` and `install_dir` end up using relative paths as well.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (31 preceding siblings ...)
  2021-01-24 17:52 ` ackalker
@ 2021-01-24 17:52 ` ackalker
  2021-01-24 17:53 ` ackalker
                   ` (31 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:52 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 901 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which supplies the rules for the substitutions inside the JSON files needs to be given an _absolute_ path for the substitutions to make any sense. Because `qemu_datadir` is relative, both `configuration` and `install_dir` end up using relative paths as well.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (32 preceding siblings ...)
  2021-01-24 17:52 ` ackalker
@ 2021-01-24 17:53 ` ackalker
  2021-01-24 17:54 ` ackalker
                   ` (30 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:53 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 916 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which supplies the rules for the substitutions of `@DATADIR@` inside the JSON files needs to be given an _absolute_ path for the substitutions to make any sense. Because `qemu_datadir` is relative, both `configuration` and `install_dir` end up using relative paths as well.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (33 preceding siblings ...)
  2021-01-24 17:53 ` ackalker
@ 2021-01-24 17:54 ` ackalker
  2021-01-24 17:56 ` ackalker
                   ` (29 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:54 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 914 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which supplies the rule for the substitution of `@DATADIR@` inside the JSON files needs to be given an _absolute_ path for the substitutions to make any sense. Because `qemu_datadir` is relative, both `configuration` and `install_dir` end up using relative paths as well.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (34 preceding siblings ...)
  2021-01-24 17:54 ` ackalker
@ 2021-01-24 17:56 ` ackalker
  2021-01-24 17:56 ` ackalker
                   ` (28 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:56 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 899 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which supplies the rule for the substitution of `@DATADIR@` inside the JSON files needs to be given an _absolute_ path for the file to be valid. Because `qemu_datadir` is relative, both `configuration` and `install_dir` end up using relative paths as well.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (35 preceding siblings ...)
  2021-01-24 17:56 ` ackalker
@ 2021-01-24 17:56 ` ackalker
  2021-01-24 17:58 ` ackalker
                   ` (27 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:56 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 901 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to the value of `prefix`, but the `configuration` keyword argument which supplies the rule for the substitution of `@DATADIR@` inside the JSON files needs to be given an _absolute_ path for the files to be usable. Because `qemu_datadir` is relative, both `configuration` and `install_dir` end up using relative paths as well.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (36 preceding siblings ...)
  2021-01-24 17:56 ` ackalker
@ 2021-01-24 17:58 ` ackalker
  2021-01-24 18:01 ` ackalker
                   ` (26 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 17:58 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 902 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to, the value of `prefix`, but the `configuration` keyword argument which supplies the rule for the substitution of `@DATADIR@` inside the JSON files needs to be given an _absolute_ path for the files to be usable. Because `qemu_datadir` is relative, both `configuration` and `install_dir` end up using relative paths as well.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (37 preceding siblings ...)
  2021-01-24 17:58 ` ackalker
@ 2021-01-24 18:01 ` ackalker
  2021-01-24 18:02 ` ackalker
                   ` (25 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 18:01 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1038 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to, the value of `prefix`, but the `configuration` keyword argument which supplies the rule for the substitution of `@DATADIR@` inside the JSON files needs to be given an _absolute_ path for the files to be usable. Because `datadir` is not configured, it defaults to `share` which is a relative path. `qemu_datadir` is derived from this value, so becomes a relative path as well, and so both `configuration` and `install_dir` end up using relative paths as well.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (38 preceding siblings ...)
  2021-01-24 18:01 ` ackalker
@ 2021-01-24 18:02 ` ackalker
  2021-01-24 18:10 ` ackalker
                   ` (24 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 18:02 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1053 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
Some more digging: the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file) states that its `install_dir` keyword argument expects a subdirectory, i.e. a path _relative_ to, the value of `prefix`, but the `configuration` keyword argument which supplies the rule for the substitution of `@DATADIR@` inside the JSON files needs to be given an _absolute_ path for the files to be usable. Because `datadir` is not configured, it defaults to `share` which is a relative path. `qemu_datadir` is derived from the value of `datadir`, so it becomes a relative path as well, and so both `configuration` and `install_dir` end up using relative paths as well.

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (39 preceding siblings ...)
  2021-01-24 18:02 ` ackalker
@ 2021-01-24 18:10 ` ackalker
  2021-01-25  2:21 ` ackalker
                   ` (23 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-24 18:10 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 972 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
From the commits I mentioned before and some more digging in the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), I found:

* `install_dir` expects a subdirectory _relative_ to `prefix`.
* `configuration` substitution rule for `DATADIR` inside the JSON files needs _absolute_ path for the files to be usable.
* As `datadir` is not configured, it defaults to `share` which is a relative path.
* `qemu_datadir` is derived from the value of `datadir`, so it ends up a relative path as well
* `configuration` and `install_dir` end up using relative paths also

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (40 preceding siblings ...)
  2021-01-24 18:10 ` ackalker
@ 2021-01-25  2:21 ` ackalker
  2021-01-25  2:23 ` ackalker
                   ` (22 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:21 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 808 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

The problem appears to be: `configure_file()` needs to be supplied both the absolute path to the firmware blobs for substitution into the descriptor files _and_ the relative path to qemu data directory for installation. Catch-22.


^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (41 preceding siblings ...)
  2021-01-25  2:21 ` ackalker
@ 2021-01-25  2:23 ` ackalker
  2021-01-25  2:23 ` ackalker
                   ` (21 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:23 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 917 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

The problem appears to be: `configure_file()` needs to be supplied both the absolute path to the firmware blobs for substitution into the descriptor files _and_ the relative path to qemu data directory for installation, see the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file). Catch-22.


^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (42 preceding siblings ...)
  2021-01-25  2:23 ` ackalker
@ 2021-01-25  2:23 ` ackalker
  2021-01-25  2:40 ` ackalker
                   ` (20 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:23 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 972 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766400380

Comment:
From the commits I mentioned before and some more digging in the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), I found:

* `install_dir` expects a subdirectory _relative_ to `prefix`.
* `configuration` substitution rule for `DATADIR` inside the JSON files needs _absolute_ path for the files to be usable.
* As `datadir` is not configured, it defaults to `share` which is a relative path.
* `qemu_datadir` is derived from the value of `datadir`, so it ends up a relative path as well
* `configuration` and `install_dir` end up using relative paths also

The way things are curently setup in Qemu's build system, setting `datadir` to an absolute path would also make `qemu_datadir` absolute, but I don't know if this violates the requirements for `configure_file()`.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (43 preceding siblings ...)
  2021-01-25  2:23 ` ackalker
@ 2021-01-25  2:40 ` ackalker
  2021-01-25  2:43 ` ackalker
                   ` (19 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:40 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1270 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file).: `configure_file()`, it becomes clear that there is a problem here:

- `configuration`: The substitution made in the input JSON file needs to be done using an _absolute_ path
- `install_dir`: the installation directory for the generated files needs to be a _relative_ path

As the commit message indicates: using an absolute path for `qemu_datadir` may result in Meson prepending an extra copy of the prefix to the installation directory when cross-compiling, and using a relative path for `qemu_datadir` results in wrong (relative) path in the generated files. A nice Catch-22.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (44 preceding siblings ...)
  2021-01-25  2:40 ` ackalker
@ 2021-01-25  2:43 ` ackalker
  2021-01-25  2:45 ` ackalker
                   ` (18 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:43 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1323 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), it becomes clear that there is a problem here:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a _relative_ path

As the commit message indicates: using an absolute path for `qemu_datadir` appears to result in Meson prepending an extra copy of the prefix to the installation directory when cross-compiling, and using a relative path for `qemu_datadir` results in wrong (relative) path in the generated files. A nice Catch-22.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (45 preceding siblings ...)
  2021-01-25  2:43 ` ackalker
@ 2021-01-25  2:45 ` ackalker
  2021-01-25  2:49 ` ackalker
                   ` (17 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:45 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1321 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), it becomes clear that there is a problem here:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a _relative_ path

As the commit message indicates: using an absolute path for `qemu_datadir` appears to result in Meson prepending an extra copy of the prefix to the installation directory when cross-compiling, and using a relative path for `qemu_datadir` results in wrong (relative) path in the generated files.

A Catch-22.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (46 preceding siblings ...)
  2021-01-25  2:45 ` ackalker
@ 2021-01-25  2:49 ` ackalker
  2021-01-25  2:51 ` ackalker
                   ` (16 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:49 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1095 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), it becomes clear that there is a problem here:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a _relative_ path

Trying to use the same variable `qemu_datadir` for both, we end up with a Catch-22.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (47 preceding siblings ...)
  2021-01-25  2:49 ` ackalker
@ 2021-01-25  2:51 ` ackalker
  2021-01-25  2:52 ` ackalker
                   ` (15 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:51 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1291 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), it becomes clear that there is a problem here:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a _relative_ path

Trying to use the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute path, then cross-compiling in a Windows environment may cause Meson to add a redundant prefix, and when its value is a relative path, the resulting files don't work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (48 preceding siblings ...)
  2021-01-25  2:51 ` ackalker
@ 2021-01-25  2:52 ` ackalker
  2021-01-25  2:55 ` ackalker
                   ` (14 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:52 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1311 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), it becomes clear that there is a problem here:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a _relative_ path

Trying to use the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute path, then cross-compiling in a Windows environment may cause Meson to add a redundant prefix, and when its value is a relative path, the resulting firmware descriptor files don't work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (49 preceding siblings ...)
  2021-01-25  2:52 ` ackalker
@ 2021-01-25  2:55 ` ackalker
  2021-01-25  2:59 ` ackalker
                   ` (13 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:55 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1294 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), it becomes clear that there is a problem here:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a _relative_ path

Trying to use the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute Windows path, then cross-compiling may cause Meson to add a redundant prefix, and when its value is a relative path, the resulting firmware descriptor files don't work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (50 preceding siblings ...)
  2021-01-25  2:55 ` ackalker
@ 2021-01-25  2:59 ` ackalker
  2021-01-25  3:02 ` ackalker
                   ` (12 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  2:59 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1297 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), it becomes clear that there is a problem here:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a _relative_ path

Trying to use the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute Windows path, then Meson may try to add a redundant prefix when cross-compiling, and when its value is a relative path, the resulting firmware descriptor files don't work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (51 preceding siblings ...)
  2021-01-25  2:59 ` ackalker
@ 2021-01-25  3:02 ` ackalker
  2021-01-25  3:04 ` ackalker
                   ` (11 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  3:02 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1291 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), the problem becomes obvious:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a path _relative_ to `prefix`

Trying to use the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute Windows path, then Meson may try to add a redundant prefix when cross-compiling, and when its value is a relative path, the resulting firmware descriptor files don't work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (52 preceding siblings ...)
  2021-01-25  3:02 ` ackalker
@ 2021-01-25  3:04 ` ackalker
  2021-01-25  3:04 ` ackalker
                   ` (10 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  3:04 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1333 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), the problem becomes obvious:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a path _relative_ to `prefix`

Trying to use the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute Windows path, then Meson may not recognize it as such when cross-compiling, and may try to add a redundant `prefix` to it, and when its value is a relative path, the resulting firmware descriptor files don't work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (53 preceding siblings ...)
  2021-01-25  3:04 ` ackalker
@ 2021-01-25  3:04 ` ackalker
  2021-01-25  3:05 ` ackalker
                   ` (9 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  3:04 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1350 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), the problem becomes obvious:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a path _relative_ to `prefix`

Trying to use variations using the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute Windows path, then Meson may not recognize it as such when cross-compiling, and may try to add a redundant `prefix` to it, and when its value is a relative path, the resulting firmware descriptor files don't work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (54 preceding siblings ...)
  2021-01-25  3:04 ` ackalker
@ 2021-01-25  3:05 ` ackalker
  2021-01-25  3:07 ` ackalker
                   ` (8 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  3:05 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1354 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), the problem becomes obvious:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a path _relative_ to `prefix`

Trying to use variations using the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute Windows path, then Meson may not recognize it as absolute when cross-compiling, and may try to add a redundant `prefix` to it, and when its value is a relative path, the resulting firmware descriptor files don't work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (55 preceding siblings ...)
  2021-01-25  3:05 ` ackalker
@ 2021-01-25  3:07 ` ackalker
  2021-01-25  3:07 ` ackalker
                   ` (7 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  3:07 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1349 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), the problem becomes obvious:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a path _relative_ to `prefix`

Trying to use variations using the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute Windows path, then Meson may not recognize it as absolute when cross-compiling and may try to add a redundant `prefix` to it. When its value is a relative path, the resulting firmware descriptor files don't work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (56 preceding siblings ...)
  2021-01-25  3:07 ` ackalker
@ 2021-01-25  3:07 ` ackalker
  2021-01-25  3:10 ` ackalker
                   ` (6 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  3:07 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1384 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), the problem becomes obvious:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a path _relative_ to `prefix`

Trying to use variations using the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute Windows path, then Meson may not recognize it as absolute when cross-compiling and may try to add a redundant `prefix` to it. When its value is a relative path, the resulting firmware descriptor files will contain relative paths and will not work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (57 preceding siblings ...)
  2021-01-25  3:07 ` ackalker
@ 2021-01-25  3:10 ` ackalker
  2021-01-25  7:04 ` ackalker
                   ` (5 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  3:10 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 1372 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766370106

Comment:
I believe that the following 2 commits to be relevant to the current issue: qemu/qemu@6aae2a2e0e3c43e479f889389074cda8bef3a580, qemu/qemu@ab4c0996f80d43d1fc28c6e76f4ecb27423a7e29.

In particular, the commit message of the 2nd commit seems relevant:

> When cross-compiling, by default qemu_datadir is 'c:\Program
Files\QEMU', which is not recognized as being an absolute path, and
meson will end up adding the prefix again.

From the [documentation for `configure_file()`](https://mesonbuild.com/Reference-manual.html#configure_file), the problem becomes obvious:

- `configuration` describes the substitutions to be made in the input file. In this case `@DATADIR@` needs to be substituted with an _absolute_ path to a firmware blob
- `install_dir` is the installation directory for the generated file, it needs to be a path _relative_ to `prefix`

Trying to use variations using the same variable `qemu_datadir` for both, we end up with a Catch-22: when its value is an absolute Windows path, then Meson doesn't recognize it as absolute when cross-compiling, and it adds a redundant `prefix`. When its value is a relative path, the resulting firmware descriptor files will contain relative paths and will not work.

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (58 preceding siblings ...)
  2021-01-25  3:10 ` ackalker
@ 2021-01-25  7:04 ` ackalker
  2021-01-25  7:08 ` ackalker
                   ` (4 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  7:04 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 242 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766596907

Comment:
I've created an upstream bug report about this issue: https://bugs.launchpad.net/qemu/+bug/1913012

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (59 preceding siblings ...)
  2021-01-25  7:04 ` ackalker
@ 2021-01-25  7:08 ` ackalker
  2021-01-25  7:09 ` ackalker
                   ` (3 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  7:08 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 511 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766596907

Comment:
I've tested passing `--datadir=<absolute path>` to `configure` in a Docker test build, but that doesn't appear to work: the `filename` attributes in the descriptor files do get the full path, but they are missing the initial `/`.
I believe this needs to be fixed upstream, so I have created a bug report about this issue: https://bugs.launchpad.net/qemu/+bug/1913012

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (60 preceding siblings ...)
  2021-01-25  7:08 ` ackalker
@ 2021-01-25  7:09 ` ackalker
  2021-01-25  7:10 ` ackalker
                   ` (2 subsequent siblings)
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  7:09 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 559 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766596907

Comment:
I've tested passing `--datadir=<absolute path>` to `configure` in a Docker test build, but that doesn't appear to work: the `filename` attributes in the descriptor files do get the full path, but they are missing the initial `/`.
I believe this needs to be fixed upstream, so I have created a bug report about this issue (with full testing steps for anyone interested): https://bugs.launchpad.net/qemu/+bug/1913012

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (61 preceding siblings ...)
  2021-01-25  7:09 ` ackalker
@ 2021-01-25  7:10 ` ackalker
  2021-01-26  2:57 ` ackalker
  2021-01-26  2:57 ` ackalker
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-25  7:10 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 592 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766596907

Comment:
I've tested passing `--datadir=<absolute path>` to `configure` in a Docker test build, but that doesn't appear to work: the `filename` attributes in the descriptor files do get the full path, but they are missing the initial `/`, so the paths are still relative.
I believe this needs to be fixed upstream, so I have created a bug report about this issue (with full testing steps for anyone interested): https://bugs.launchpad.net/qemu/+bug/1913012

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (62 preceding siblings ...)
  2021-01-25  7:10 ` ackalker
@ 2021-01-26  2:57 ` ackalker
  2021-01-26  2:57 ` ackalker
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-26  2:57 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 595 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766596907

Comment:
I've tested passing `--datadir=<absolute path>` to `configure` in a Docker test build, but that doesn't appear to work: the `filename` attributes in the descriptor files do get the full path, but they are missing the initial prefix, so the paths are still relative.
I believe this needs to be fixed upstream, so I have created a bug report about this issue (with full testing steps for anyone interested): https://bugs.launchpad.net/qemu/+bug/1913012

^ permalink raw reply	[flat|nested] 66+ messages in thread

* Re: qemu: datadir is not set, uefi support is broken
  2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
                   ` (63 preceding siblings ...)
  2021-01-26  2:57 ` ackalker
@ 2021-01-26  2:57 ` ackalker
  64 siblings, 0 replies; 66+ messages in thread
From: ackalker @ 2021-01-26  2:57 UTC (permalink / raw)
  To: ml

[-- Attachment #1: Type: text/plain, Size: 592 bytes --]

New comment by ackalker on void-packages repository

https://github.com/void-linux/void-packages/issues/27965#issuecomment-766596907

Comment:
I've tested passing `--datadir=<absolute path>` to `configure` in a Docker test build, but that doesn't appear to work: the `filename` attributes in the descriptor files do get the full path, but they are missing the initial `/`, so the paths are still relative.
I believe this needs to be fixed upstream, so I have created a bug report about this issue (with full testing steps for anyone interested): https://bugs.launchpad.net/qemu/+bug/1913012

^ permalink raw reply	[flat|nested] 66+ messages in thread

end of thread, other threads:[~2021-01-26  2:57 UTC | newest]

Thread overview: 66+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-01-16 10:40 [ISSUE] qemu: datadir is not set, uefi support is broken gspe
2021-01-19 21:22 ` ericonr
2021-01-19 22:51 ` gspe
2021-01-19 22:55 ` ericonr
2021-01-20 17:55 ` gspe
2021-01-20 17:55 ` gspe
2021-01-21  1:41 ` ericonr
2021-01-21  1:42 ` ericonr
2021-01-21 21:02 ` gspe
2021-01-21 21:10 ` ericonr
2021-01-21 21:15 ` gspe
2021-01-21 21:19 ` ericonr
2021-01-21 21:21 ` gspe
2021-01-21 21:22 ` gspe
2021-01-21 21:23 ` gspe
2021-01-21 21:24 ` gspe
2021-01-21 21:28 ` ericonr
2021-01-24 15:14 ` ackalker
2021-01-24 15:17 ` ackalker
2021-01-24 15:19 ` ackalker
2021-01-24 15:21 ` ackalker
2021-01-24 15:28 ` ackalker
2021-01-24 15:42 ` ackalker
2021-01-24 15:46 ` ackalker
2021-01-24 15:48 ` ackalker
2021-01-24 17:39 ` ackalker
2021-01-24 17:39 ` ackalker
2021-01-24 17:45 ` ackalker
2021-01-24 17:46 ` ackalker
2021-01-24 17:47 ` ackalker
2021-01-24 17:48 ` ackalker
2021-01-24 17:51 ` ackalker
2021-01-24 17:52 ` ackalker
2021-01-24 17:52 ` ackalker
2021-01-24 17:53 ` ackalker
2021-01-24 17:54 ` ackalker
2021-01-24 17:56 ` ackalker
2021-01-24 17:56 ` ackalker
2021-01-24 17:58 ` ackalker
2021-01-24 18:01 ` ackalker
2021-01-24 18:02 ` ackalker
2021-01-24 18:10 ` ackalker
2021-01-25  2:21 ` ackalker
2021-01-25  2:23 ` ackalker
2021-01-25  2:23 ` ackalker
2021-01-25  2:40 ` ackalker
2021-01-25  2:43 ` ackalker
2021-01-25  2:45 ` ackalker
2021-01-25  2:49 ` ackalker
2021-01-25  2:51 ` ackalker
2021-01-25  2:52 ` ackalker
2021-01-25  2:55 ` ackalker
2021-01-25  2:59 ` ackalker
2021-01-25  3:02 ` ackalker
2021-01-25  3:04 ` ackalker
2021-01-25  3:04 ` ackalker
2021-01-25  3:05 ` ackalker
2021-01-25  3:07 ` ackalker
2021-01-25  3:07 ` ackalker
2021-01-25  3:10 ` ackalker
2021-01-25  7:04 ` ackalker
2021-01-25  7:08 ` ackalker
2021-01-25  7:09 ` ackalker
2021-01-25  7:10 ` ackalker
2021-01-26  2:57 ` ackalker
2021-01-26  2:57 ` ackalker

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).