Re: [9front] T60 regression?

Re: [9front] T60 regression?

[cinap_lenrek]

excellent. thanks for the summary.

its not clear to me what change could have caused this yet.

i'd try to boot with *acpi= first if that hasnt been tried
already. maybe its not really a regression but the mp tables
where always broken on T60?

--
cinap

Re: [9front] T60 regression?

[Stanley Lieber]

> maybe its not really a regression but the mp tables
> where always broken on T60?

IIRC, mp tables were broken on *60 and *61. Way back when, I had to use the *mptables= (or whatever it was) hack. Again IIRC, this hack was removed when we got *acpi=.

sl

Re: [9front] T60 regression?

[istvan bak]

The below changeset reproduces this (and later ones too), and the ones
before it don't. I tested on amd64, with pc64 kernel it reproduces,
with pc kernel it doesn't. (I am not gln34, and my machine isn't a
T60, nor a laptop.)

changeset:   6786:0a937742a16b
parent:      6784:3b33f587ad27
user:        cinap_lenrek@felloff.net
date:        Sat Oct 06 00:26:07 2018 +0200
summary:     pc, pc64: increase CONFADDR buffer from 4K to 24K

sysinfo:
http://sysinfo.9front.org/src/256/body

The symptoms are the same (hang at lapic clock, resume booting,
garbage on screen). *nomp=1 hangs it forever, *acpi=0 doesn't affect
it. sorry I know little, don't know how to help more.

paperwing

Re: [9front] T60 regression?

[istvan bak]

I can confirm the patch works, thank you.

Re: [9front] T60 regression?

[cinap_lenrek]

ok, i think i got it. the mem.h dependency was not expressed
correctly for apbootstrap.h in pc64. there was a "apboostrap.$O: mem.h"
dependency but this is wrong as apoostrap.$O is an intermediate file
that never gets referenced in the dependency graph. the real dependency
needs to be on apboostrap.h.

changeset:   6986:b048db97b5ea
tag:         tip
user:        cinap_lenrek@felloff.net
date:        Fri Jan 04 02:51:29 2019 +0100
summary:     pc, pc64: properly track dependencies for mem.h on autogenerated apbootstrap.h and reboot.h targets

diff -r 8c50688cb280 -r b048db97b5ea sys/src/9/pc/mkfile
--- a/sys/src/9/pc/mkfile	Wed Jan 02 10:26:38 2019 +0100
+++ b/sys/src/9/pc/mkfile	Fri Jan 04 02:51:29 2019 +0100
@@ -8,8 +8,9 @@
 p=9
 
 # must match mem.h
+KTZERO=0xF0100020
 APBOOTSTRAP=0xF0007000
-KTZERO=0xF0100020
+REBOOTADDR=0x11000
 
 DEVS=`{rc ../port/mkdevlist $CONF}
 
@@ -143,15 +144,15 @@
 		sed -e 's/^[0-9a-f]+ //' -e 's/ ([0-9a-f][0-9a-f])/0x\1,/g'
 	 echo '};'} > init.h
 
-reboot.h:D:	rebootcode.s
+reboot.h:D:	rebootcode.s mem.h
 	$AS rebootcode.s
-	$LD -l -s -T0x11000 -R4 -o reboot.out rebootcode.$O
+	$LD -l -s -T$REBOOTADDR -R4 -o reboot.out rebootcode.$O
 	{echo 'uchar rebootcode[]={'
 	 xd -1x reboot.out |
 		sed -e '1,2d' -e 's/^[0-9a-f]+ //' -e 's/ ([0-9a-f][0-9a-f])/0x\1,/g'
 	 echo '};'} > reboot.h
 
-apbootstrap.h:	apbootstrap.s mem.h
+apbootstrap.h:D:	apbootstrap.s mem.h
 	$AS $prereq
 	$LD -o apbootstrap.out -T$APBOOTSTRAP -R4 -l -s apbootstrap.$O
 	{echo 'uchar apbootstrap[]={'
diff -r 8c50688cb280 -r b048db97b5ea sys/src/9/pc64/mkfile
--- a/sys/src/9/pc64/mkfile	Wed Jan 02 10:26:38 2019 +0100
+++ b/sys/src/9/pc64/mkfile	Fri Jan 04 02:51:29 2019 +0100
@@ -5,6 +5,7 @@
 </$objtype/mkfile
 p=9
 
+# must match mem.h
 KTZERO=0xffffffff80110000
 APBOOTSTRAP=0xffffffff80007000
 REBOOTADDR=0x11000
@@ -99,7 +100,7 @@
 <../port/portmkfile
 <|../port/mkbootrules $CONF
 
-l.$O apbootstrap.$O:		mem.h
+l.$O:				mem.h
 
 $ETHER: 			../port/etherif.h ../port/netif.h
 $AUDIO:				../port/audioif.h
@@ -144,7 +145,7 @@
 		sed -e 's/^[0-9a-f]+ //' -e 's/ ([0-9a-f][0-9a-f])/0x\1,/g'
 	 echo '};'} > init.h
 
-reboot.h:	rebootcode.s
+reboot.h:D:	rebootcode.s mem.h
 	$AS rebootcode.s
 	$LD -l -R1 -s -o reboot.out -T$REBOOTADDR rebootcode.$O
 	{echo 'uchar rebootcode[]={'
@@ -153,7 +154,7 @@
 	 sed -e 's/^[0-9a-f]+ //' -e 's/ ([0-9a-f][0-9a-f])/0x\1,/g'
 	 echo '};'} > $target
 
-apbootstrap.h:	apbootstrap.s
+apbootstrap.h:D:	apbootstrap.s mem.h
 	$AS apbootstrap.s
 	$LD -l -R1 -s -o apbootstrap.out -T$APBOOTSTRAP apbootstrap.$O
 	{echo 'uchar apbootstrap[]={'

Re: [9front] T60 regression?

[cinap_lenrek]

hm, could this be a build issue? the link address of apbootstrap
changed, but the apboostrap.h target doesnt account for changes
of the mkfile itself. try mk clean so that apbootstrap.h gets
rebuild.

-APBOOTSTRAP=0xffffffff80003000
+APBOOTSTRAP=0xffffffff80007000

if this was the issue, then i think the fix would be to add the
mkfile as an dependency to reboot.h and apbootstrap.h targets.

--
cinap

Re: [9front] T60 regression?

[Kurt H Maier]

On Thu, Jan 03, 2019 at 09:31:31AM +0100, cinap_lenrek@felloff.net wrote:
> a few days ago, gln34 reported that after updating that
> boot would fail on his ThinkPad T60:

I don't have a T60 at the moment, but here is gln34's sysinfo:
http://sysinfo.9front.org/src/250/body

He reported the system booted normally with *nomp=1.  I got him to take
a photo of a bad boot (without *nomp=1): 
This is where it hangs:  https://i.imgur.com/CM1zpEN.jpg
This is the corrupt rio: https://i.imgur.com/85wTU1h.jpg
Good boot with *nomp=1: https://i.imgur.com/5wn4Kss.jpg

Sorry if you already saw all this; I just wanted to get all the reported
information in one place.

I'm working on getting my T60 back but I don't know how long it will
take.

khm