Short-term todo list - 18 Jul

Short-term todo list - 18 Jul

[Rich Felker]

Aside from the big development items left in this release cycle
(adding locale functionality and IDN), there are a few more short-term
tasks I want to get done in the next day or two based on open mailing
list threads and other discussions. The ones I'm aware of right now
are:

- Merging or1k
- Finishing discussion of issetugid and hopefully merging
- Reviewing glob_pattern_p patch
- Figuring out what to do with atomic asm constaints & volatile
- Fixing microblaze stat structure (& getting qemu fixed?)
- Addressing mips stat structure dev_t endianness issue

Is there anything else missing from this list?

Rich

Re: Short-term todo list - 18 Jul

[Szabolcs Nagy]

* Rich Felker <dalias@libc.org> [2014-07-18 01:15:58 -0400]:
> - Merging or1k
> - Finishing discussion of issetugid and hopefully merging
> - Reviewing glob_pattern_p patch
> - Figuring out what to do with atomic asm constaints & volatile
> - Fixing microblaze stat structure (& getting qemu fixed?)
> - Addressing mips stat structure dev_t endianness issue
> 
> Is there anything else missing from this list?
> 

probably not in this release but further todo before i forget:

- i have patch with linux 3.15 updates (mostly SYS_renameat2)
can commit it after or1k merge

- i have an adjtimex.c for x32, but it is broken because kernel
timex struct has 64bit values while userspace struct is 32bit
and it has even padding for future extensions which cannot be
handled if they may add either kernel_long or int members

- with yet another soft-float arch the float usage in malloc
should be probably revisited (fp code is faster for x86_64,
but bigger and slower for embedded soft-float things)

Re: Short-term todo list - 18 Jul

[Rich Felker]

On Fri, Jul 18, 2014 at 01:15:58AM -0400, Rich Felker wrote:
> Aside from the big development items left in this release cycle
> (adding locale functionality and IDN), there are a few more short-term
> tasks I want to get done in the next day or two based on open mailing
> list threads and other discussions. The ones I'm aware of right now
> are:
> 
> - Merging or1k

Done.

> - Finishing discussion of issetugid and hopefully merging

Discussion done. Just want to make a good commit message explaining
the rationale and I think it's ready to commit.

> - Reviewing glob_pattern_p patch

Done. Does anyone object to adding this or think it should be done
differently if added?

> - Figuring out what to do with atomic asm constaints & volatile

Almost done. I still want to add __volatile__ to all the archs that
are missing it and add the memory constraints for microblaze.

Also a_store seems to be broken on a number of archs... I'm trying to
figure out what to do with this too.

> - Fixing microblaze stat structure (& getting qemu fixed?)

Done. (Minus getting qemu fixed; anyone willing to volunteer to follow
up on that?)

> - Addressing mips stat structure dev_t endianness issue

I think this should be hidden away in arch/mips/syscall_arch.h but the
way the macros are done makes it ugly and difficult to add. I might
just flatten all the __syscallN functions.

Rich

Re: Short-term todo list - 18 Jul

[Rich Felker]

On Sat, Jul 19, 2014 at 05:01:01PM -0400, Rich Felker wrote:
> > - Finishing discussion of issetugid and hopefully merging
> 
> Discussion done. Just want to make a good commit message explaining
> the rationale and I think it's ready to commit.

Done.

> > - Figuring out what to do with atomic asm constaints & volatile
> 
> Almost done. I still want to add __volatile__ to all the archs that
> are missing it and add the memory constraints for microblaze.

These minor details are still open.

> Also a_store seems to be broken on a number of archs... I'm trying to
> figure out what to do with this too.

I think all versions of a_store except possibly sh are correct now.
Waiting on feedback from Bobby Bingham (sh port author) on it since I
don't understand the logic of the current code.

> > - Addressing mips stat structure dev_t endianness issue
> 
> I think this should be hidden away in arch/mips/syscall_arch.h but the
> way the macros are done makes it ugly and difficult to add. I might
> just flatten all the __syscallN functions.

Still pending...

Rich

Re: Short-term todo list - 18 Jul

[Rich Felker]

On Sat, Jul 19, 2014 at 09:50:09PM -0400, Rich Felker wrote:
> > > - Addressing mips stat structure dev_t endianness issue
> > 
> > I think this should be hidden away in arch/mips/syscall_arch.h but the
> > way the macros are done makes it ugly and difficult to add. I might
> > just flatten all the __syscallN functions.
> 
> Still pending...

And now done.

Rich