Open issues at end of April

Open issues at end of April

[Rich Felker]

Currently pending things I haven't gotten to yet, or which have open
discussion still, seem to be:

- Request for renaming dynamic linker's _start.
- max_align_t
- Changes for no-legacy-syscall archs
- __xmknod and __sysv_signal ABI symbols
- fmtmsg

There are also several resolver-related topics which are going to be
addressed in the resolver overhaul next month (May).

I believe the ARM build issues have been addressed; let me know if
anything remains open there.

I'm planning to release 1.1.1 and 1.0.2 right away with some important
fixes (mainly the strstr-family stuff), and work on the above items as
part of 1.1.2. Please remind me if there are other pending things I'm
forgetting.

Rich

Re: Open issues at end of April

[Timo Teras]

On Wed, 30 Apr 2014 20:40:58 -0400
Rich Felker <dalias@libc.org> wrote:

> Currently pending things I haven't gotten to yet, or which have open
> discussion still, seem to be:
> 
> - Request for renaming dynamic linker's _start.
> - max_align_t
> - Changes for no-legacy-syscall archs
> - __xmknod and __sysv_signal ABI symbols
> - fmtmsg
> 
> There are also several resolver-related topics which are going to be
> addressed in the resolver overhaul next month (May).
> 
> I believe the ARM build issues have been addressed; let me know if
> anything remains open there.
> 
> I'm planning to release 1.1.1 and 1.0.2 right away with some important
> fixes (mainly the strstr-family stuff), and work on the above items as
> part of 1.1.2. Please remind me if there are other pending things I'm
> forgetting.

1. I would need to have cp437 and cp850 in iconv. Otherwise mtools
cannot accesss fat filesystems, meaning syslinux is broke. I know iconv
overhaul is scheduled for 1.1.4 with hopefully getting the table
generation code published. But I think these few legacy codepages
should be added as a bug fix.

2. Possibly also sysconf() fixes: mapping table 0 to return value -1,
and fixing LONG_BIT should be done.

3. I also think having CFI for at least syscall_cpi should be done.
Debugging without that is a pain.

4. Other items in my wish list is "sendmmsg() and recvmmsg()" but as
discussed it's tricky.

5. And I also had the question if for NEEDED dependencies constructors
should be topologically sorted.

1-3 being 'high' priority.

- Timo

Re: Open issues at end of April

[Timo Teras]

On Thu, 1 May 2014 08:17:58 +0300
Timo Teras <timo.teras@iki.fi> wrote:

> On Wed, 30 Apr 2014 20:40:58 -0400
> Rich Felker <dalias@libc.org> wrote:
> 
> > Currently pending things I haven't gotten to yet, or which have open
> > discussion still, seem to be:
> > 
> > - Request for renaming dynamic linker's _start.
> > - max_align_t
> > - Changes for no-legacy-syscall archs
> > - __xmknod and __sysv_signal ABI symbols
> > - fmtmsg
> > 
> > There are also several resolver-related topics which are going to be
> > addressed in the resolver overhaul next month (May).
> > 
> > I believe the ARM build issues have been addressed; let me know if
> > anything remains open there.
> > 
> > I'm planning to release 1.1.1 and 1.0.2 right away with some
> > important fixes (mainly the strstr-family stuff), and work on the
> > above items as part of 1.1.2. Please remind me if there are other
> > pending things I'm forgetting.
> 
> 1. I would need to have cp437 and cp850 in iconv. Otherwise mtools
> cannot accesss fat filesystems, meaning syslinux is broke. I know
> iconv overhaul is scheduled for 1.1.4 with hopefully getting the table
> generation code published. But I think these few legacy codepages
> should be added as a bug fix.
> 
> 2. Possibly also sysconf() fixes: mapping table 0 to return value -1,
> and fixing LONG_BIT should be done.
> 
> 3. I also think having CFI for at least syscall_cpi should be done.
> Debugging without that is a pain.
> 
> 4. Other items in my wish list is "sendmmsg() and recvmmsg()" but as
> discussed it's tricky.
> 
> 5. And I also had the question if for NEEDED dependencies constructors
> should be topologically sorted.
> 
> 1-3 being 'high' priority.

6. Implementing if_nameindex() and getifaddrs() properly. Usable Netlink
version exists (which might still need some fine tuning):
http://git.alpinelinux.org/cgit/aports/plain/main/musl/1002-reimplement-if_nameindex-and-getifaddrs-using-netlin.patch
Can wait until 1.1.2.

7. To go with 1.1.2 DNS overhaul: adding dns record parsing function:
http://git.alpinelinux.org/cgit/aports/plain/main/musl/1001-add-basic-dns-record-parsing-functions.patch

- Timo

Re: Open issues at end of April

[Natanael Copa]

On Wed, 30 Apr 2014 20:40:58 -0400
Rich Felker <dalias@libc.org> wrote:

> Currently pending things I haven't gotten to yet, or which have open
> discussion still, seem to be:
> 
> - Request for renaming dynamic linker's _start.
> - max_align_t
> - Changes for no-legacy-syscall archs
> - __xmknod and __sysv_signal ABI symbols
> - fmtmsg
> 
> There are also several resolver-related topics which are going to be
> addressed in the resolver overhaul next month (May).
> 
> I believe the ARM build issues have been addressed; let me know if
> anything remains open there.
> 
> I'm planning to release 1.1.1 and 1.0.2 right away with some important
> fixes (mainly the strstr-family stuff), and work on the above items as
> part of 1.1.2. Please remind me if there are other pending things I'm
> forgetting.

getservbyname/getaddrinfo to support aliases in /etc/services.

But i don't know if there is so much to discuss there. It just needs to
be done.

> Rich

Re: Open issues at end of April

[Rich Felker]

On Thu, May 01, 2014 at 08:17:58AM +0300, Timo Teras wrote:
> 1. I would need to have cp437 and cp850 in iconv. Otherwise mtools

This one is done now.

> 2. Possibly also sysconf() fixes: mapping table 0 to return value -1,
> and fixing LONG_BIT should be done.

I did the -1 stuff, and added a distinction between -1 as "no fixed
limit" and -1 as "invalid argument" (the latter now sets EINVAL).
LONG_BIT etc. could in principle be supported (though ULONG_MAX is
nonsense) but I'd like to hold off on adding something that's rather
silly (not runtime variable) and probably shouldn't be there. Since
it's not (can can't be) runtime variable, getconf(1) should probably
just hard-code these (using limits.h) if it wants to provide them.

> 3. I also think having CFI for at least syscall_cpi should be done.
> Debugging without that is a pain.

As we discussed, let's leave this as a local patch on your side for
now. I'm still hoping (perhaps this is futile :/) to find a way to do
it that (1) doesn't uglify all the sources, and (2) automatically gets
omitted from .o files when debug symbols are turned off (since .o
files compiled from C source won't have it, it doesn't make sense for
the ones from asm source to have it).

> 4. Other items in my wish list is "sendmmsg() and recvmmsg()" but as
> discussed it's tricky.

Yes, even some new issues came up: emulating with sendmsg/recvmsg is
not valid because the timeout can't be emulated. However I think we
could emulate sendmmsg with N messages as N calls to sendmmsg with
just one message each, or some similar breakdown, such that the amount
of data that needs to be copied and patched-up is bounded.

> 5. And I also had the question if for NEEDED dependencies constructors
> should be topologically sorted.

I'd like to avoid making any guarantees about ctor/dtor order; doing
so can place nasty constraints on how some dynamic-loading situations
are handled, and only benefits broken apps/libs. I think you can get
similar behavior in a clean portable way by having init functions that
run as ctors also call (via pthread_once) the other init functions
they depend on.

Rich

Re: Open issues at end of April

[Rich Felker]

On Thu, May 01, 2014 at 08:33:08AM +0300, Timo Teras wrote:
> 6. Implementing if_nameindex() and getifaddrs() properly. Usable Netlink
> version exists (which might still need some fine tuning):
> http://git.alpinelinux.org/cgit/aports/plain/main/musl/1002-reimplement-if_nameindex-and-getifaddrs-using-netlin.patch
> Can wait until 1.1.2.

OK. In the mean time you're welcome to keep using your local patch. :)
In fact part of the new Guidelines for Distribution on the wiki was
written with this specific issue in mind:

http://wiki.musl-libc.org/wiki/Guidelines_for_Distributions

This issue will be fixed at some point in the future, hopefully in
1.1.2. I would very much like to find a solution where we get
"acceptable" behavior on BSD kernels that implement the Linux syscall
API but not netlink, however.

> 7. To go with 1.1.2 DNS overhaul: adding dns record parsing function:
> http://git.alpinelinux.org/cgit/aports/plain/main/musl/1001-add-basic-dns-record-parsing-functions.patch

Yes, this has always been part of the intent for that agenda item.

Rich

Re: Open issues at end of April

[Rich Felker]

On Thu, May 01, 2014 at 09:08:42AM +0200, Natanael Copa wrote:
> getservbyname/getaddrinfo to support aliases in /etc/services.
> 
> But i don't know if there is so much to discuss there. It just needs to
> be done.

I think this falls under resolver overhaul. For a short-term
workaround I'd recommend either changing /etc/services to use separate
lines rather than aliases or applying your own local patch. I don't
want to spend time reviewing patches for code that's about to be
highly refactored/overhauled though (and maybe you don't want to write
it either) so the first option may be the best/easiest for everybody.

Rich