klibc - Aug 2011 - klibc current state

Hello,

While some people might scuba dive, let's summarize recent
progress and state (People cc'ed have either patches in
queue or are involved):

* Patches pending review, allmost ready for klibc 
-------------------------------------------------

- mikew/drop_cap - kinit: Add drop_capabilities support
  adds the ability to kinit to allow the dropping of POSIX
  capabilities. (patch even referenced on lwn :)

- mikew/run-parts - kinit: Add run_parts()
  adds possibility to run boot scripts from kinit

- maks/fdopendir - dirent.h add fdopendir()
  initial implementation

aboves patches are for wider testing in:
http://git.kernel.org/?p=libs/klibc/klibc-queue.git


* Patches in the works:
-----------------------

- mikew/bootscripts - kinit: Add callsites to execute files
  Needs discussion about /scripts path, maybe look at was udev
  does, like /lib/kinit/ ??

- maks/switch_root : Reworks run-init into feature equiv with switch_root
  Needs repost on top of aboves queue

- hpa/stdio : Buffered stdio
  Small conflicts with missing ferror (no real blocker as current ferror
  does nothing). 

- maks/mv : Fix utils mv to work across fs
  Implemented recursively so can also serve as minimal cp.

- maks/epoll: Add initial implementation
  Need to finish test then will post.

- dash_sync : Sync up with latest git dash.
  Kbuild stumbling block of redirected stdout.
 
- tg/mkstemp : Add minimal mktemps(3) implementation
  Needed for mksh, saw no repost yet after first round.

- kumaan/ipconfig: append DHCP options length
  We may need to first implement domain-search and lease time

Either the hpa/stdio and maks/switch_init will trigger
an increment in the major version.

There exists a newly formed #klibc irc channel on freenode
and a wiki on kernel.org was asked for.

Thank you

-- 
maks

On Fri, Aug 12, 2011 at 09:14:40AM +0200, maximilian attems
wrote:
> Hello,
> 
> While some people might scuba dive, let's summarize recent
> progress and state (People cc'ed have either patches in
> queue or are involved):
Thanks for the update!

Where is klibc actually in use today?
It is nice to know a little about what any eventual contribution is used for :-)

The plan to include klibc with the kernel have stalled - seems we are missing
a killing feature to get it included...

	Sam

forgot in this first compilation 2 arch patches.

On Fri, 12 Aug 2011, maximilian attems wrote:
> Hello,
> 
> While some people might scuba dive, let's summarize recent
> progress and state (People cc'ed have either patches in
> queue or are involved):
> 
> * Patches pending review, allmost ready for klibc 
> -------------------------------------------------
> 
> - mikew/drop_cap - kinit: Add drop_capabilities support
>   adds the ability to kinit to allow the dropping of POSIX
>   capabilities. (patch even referenced on lwn :)
> 
> - mikew/run-parts - kinit: Add run_parts()
>   adds possibility to run boot scripts from kinit
> 
> - maks/fdopendir - dirent.h add fdopendir()
>   initial implementation
> 
> aboves patches are for wider testing in:
> http://git.kernel.org/?p=libs/klibc/klibc-queue.git
> 
> 
> * Patches in the works:
> -----------------------
> 
> - mikew/bootscripts - kinit: Add callsites to execute files
>   Needs discussion about /scripts path, maybe look at was udev
>   does, like /lib/kinit/ ??
> 
> - maks/switch_root : Reworks run-init into feature equiv with switch_root
>   Needs repost on top of aboves queue
> 
> - hpa/stdio : Buffered stdio
>   Small conflicts with missing ferror (no real blocker as current ferror
>   does nothing). 
> 
> - maks/mv : Fix utils mv to work across fs
>   Implemented recursively so can also serve as minimal cp.
> 
> - maks/epoll: Add initial implementation
>   Need to finish test then will post.
> 
> - dash_sync : Sync up with latest git dash.
>   Kbuild stumbling block of redirected stdout.
>  
> - tg/mkstemp : Add minimal mktemps(3) implementation
>   Needed for mksh, saw no repost yet after first round.
> 
> - kumaan/ipconfig: append DHCP options length
>   We may need to first implement domain-search and lease time
- brad/avr32 Add initial avr32 port
  No feedback yet from linux arch maintainers (wait - only 48 hours)

- tg/m68k Fix m68k syscall API and support 6-argument syscalls.
  No feedback from linux arch maintainers, needs repost.
 
> Either the hpa/stdio and maks/switch_init will trigger
> an increment in the major version.
> 
> There exists a newly formed #klibc irc channel on freenode
> and a wiki on kernel.org was asked for.
> 
> Thank you
-- 
maks

On Fri, Aug 12, 2011 at 12:14 AM, maximilian attems <max at stro.at>
wrote:
> * Patches in the works:
> -----------------------
>
> - mikew/bootscripts - kinit: Add callsites to execute files
> ?Needs discussion about /scripts path, maybe look at was udev
> ?does, like /lib/kinit/ ??
I don't know what the next steps here are, though I think there are two
issues:

- Where do we root the drop directories.  Currently /scripts, but
/lib/kinit works too.
- What callsites do we need.  Currently, I posted:
  "after-ipconfig"
  "after-mount"

Both have obvious names, which may or may not be in good taste.   The
only other "obvious" place I see there being a potential callsite
would be "before-resume".   I'm hesitant to add any callsites
before
we actually have use-cases for them, and the only one that I rely upon
at the moment is the "after-mount" one.

What are your thoughts on switching to /lib/kinit for the base
directory and iterating from there?

maximilian attems dixit:
>are unused to such minimal utilities and expect standard tools
>like more (+ cats and dogs;). On the other side and only klibc
mksh includes a sample implementation of more in shell ;-)
and isn?t _that_ much bigger than *ash. (And it deals well
with minimal environments, such as Bionic.)
>- tg/m68k Fix m68k syscall API and support 6-argument syscalls.
It?s probably incomplete ? when building mksh on m68k with
it, I noticed that some tests still fail. But it?s a good
start. Plus it needs some assembly glue since klibc reverted
back to Undefined Behaviour in C which breaks here.

No answer from the porters that I am aware of.

bye,
//mirabilos
-- 
08:05?<XTaran:#grml> mika: Does grml have an tool to read Apple
     ?    System Log (asl) files? :)
08:08?<ft:#grml> yeah. /bin/rm. ;)       08:09?<mrud:#grml> hexdump
-C
08:31?<XTaran:#grml> ft, mrud: *g*