Re: [libposix-development] Selective compilation

[Henrique A. <hd...@gm...>]

 I think bzip2 is a good one to start. It has aroung 6.000 lines of
code, doesn't use a configure script, and doesn't use extensions (some
extended operations are automatically included by GNU libc and won't
interfere with our implementation). The function list in the bzip2
binary (when compiled with glibc, this is an approximate list for us)
follows. Please choose some functions to implement. Don't worry about
fully implementing the functions, only the features needed by bzip2
need to be present. Later, I'll write unit tests so that we can
gradually complete the implementation. Use the standard to derive the
implementation. Assume that linux system calls (such as "open" or
"write") conform to POSIX 2008.

 I'm planning to add a printf parser to the repository soon. :-)

Relocation section '.rel.plt' at offset 0x858 contains 40 entries:
 Offset     Info    Type            Sym.Value  Sym. Name
0805d000  00000107 R_386_JUMP_SLOT   00000000   fileno
0805d004  00000207 R_386_JUMP_SLOT   00000000   __errno_location
0805d008  00000307 R_386_JUMP_SLOT   00000000   strerror
0805d00c  00000407 R_386_JUMP_SLOT   00000000   open64
0805d010  00000507 R_386_JUMP_SLOT   00000000   __fprintf_chk
0805d014  00000607 R_386_JUMP_SLOT   00000000   signal
0805d018  00000707 R_386_JUMP_SLOT   00000000   __gmon_start__
0805d01c  00000807 R_386_JUMP_SLOT   00000000   rewind
0805d020  00000907 R_386_JUMP_SLOT   00000000   __xstat64
0805d024  00000a07 R_386_JUMP_SLOT   00000000   getenv
0805d028  00000b07 R_386_JUMP_SLOT   00000000   strncpy
0805d02c  00000c07 R_386_JUMP_SLOT   00000000   fchown
0805d030  00000d07 R_386_JUMP_SLOT   00000000   fopen64
0805d034  00000e07 R_386_JUMP_SLOT   00000000   __libc_start_main
0805d038  00000f07 R_386_JUMP_SLOT   00000000   perror
0805d03c  00001007 R_386_JUMP_SLOT   00000000   __strcat_chk
0805d040  00001107 R_386_JUMP_SLOT   00000000   ungetc
0805d044  00001207 R_386_JUMP_SLOT   00000000   fdopen
0805d048  00001307 R_386_JUMP_SLOT   00000000   free
0805d04c  00001407 R_386_JUMP_SLOT   00000000   __lxstat64
0805d050  00001507 R_386_JUMP_SLOT   00000000   fflush
0805d054  00001607 R_386_JUMP_SLOT   00000000   __ctype_b_loc
0805d058  00001707 R_386_JUMP_SLOT   00000000   isatty
0805d05c  00001807 R_386_JUMP_SLOT   00000000   fclose
0805d060  00001907 R_386_JUMP_SLOT   00000000   utime
0805d064  00001a07 R_386_JUMP_SLOT   00000000   strlen
0805d068  00001b07 R_386_JUMP_SLOT   00000000   fgetc
0805d06c  00001c07 R_386_JUMP_SLOT   00000000   strcpy
0805d070  00001d07 R_386_JUMP_SLOT   00000000   close
0805d074  00001e07 R_386_JUMP_SLOT   00000000   fwrite
0805d078  00001f07 R_386_JUMP_SLOT   00000000   strstr
0805d07c  00002007 R_386_JUMP_SLOT   00000000   remove
0805d080  00002107 R_386_JUMP_SLOT   00000000   malloc
0805d084  00002207 R_386_JUMP_SLOT   00000000   __stack_chk_fail
0805d088  00002307 R_386_JUMP_SLOT   00000000   fputc
0805d08c  00002407 R_386_JUMP_SLOT   00000000   fread
0805d090  00002507 R_386_JUMP_SLOT   00000000   ferror
0805d094  00002607 R_386_JUMP_SLOT   00000000   strcmp
0805d098  00002707 R_386_JUMP_SLOT   00000000   exit
0805d09c  00002807 R_386_JUMP_SLOT   00000000   fchmod

2009/6/18 Henrique Almeida <hd...@gm...>:
>  I've made a list of all the packages that install files in /bin and
> /sbin in my system. Some of them are simple. Maybe bzip2 is a possible
> choice, however it uses extensions, at least in x86 (it uses the
> open64 family of functions). Maybe we can compile it without
> extensions. Also, we can get the list of all the functions imported by
> the binaries with realdelf. With the function list we're building, we
> could attribute a difficulty grade for supporting each program, and
> follow the order, starting from the easiest. :-)
>
> alsa-base
> alsa-utils
> apparmor
> bash
> binutils-static
> brltty
> bzip2
> console-setup
> coreutils
> cpio
> csh
> dash
> dbus
> debianutils
> dhcp3-client
> dmsetup
> dosfstools
> dpkg
> e2fsprogs
> ed
> fuse-utils
> grep
> grub
> gzip
> hal
> hdparm
> hostname
> ifupdown
> initscripts
> iproute
> iptables
> iputils-ping
> kbd
> klogd
> libc6
> libpam-modules
> linux-restricted-modules-common
> login
> makedev
> mktemp
> module-init-tools
> mount
> nano
> netcat-traditional
> net-tools
> ntfs-3g
> parted
> passwd
> pcmciautils
> powermgmt-base
> procps
> psmisc
> readahead
> reiserfsprogs
> sed
> sysklogd
> sysvinit-utils
> tar
> udev
> upstart
> upstart-compat-sysv
> upstart-logd
> usplash
> util-linux
> wireless-crda
> wireless-tools
> wpasupplicant
> xfsprogs
>
> 2009/6/18 Chris Forbes <ch...@fa...>:
>> I think that would be a good idea.
>> That pile of m4 code is ... large :)
>>
>> -----Original Message-----
>> From: Henrique Almeida [mailto:hd...@gm...]
>> Sent: Friday, 19 June 2009 4:18 a.m.
>> To: lib...@li...
>> Subject: Re: [libposix-development] Selective compilation
>>
>>  Maybe we should choose another application ?
>>
>> 2009/6/18 Jim Meyering <ji...@me...>:
>>> Henrique Almeida wrote:
>>>>  I'm building a new libc implementation, called libposix, with a few
>>>> volunteers. We'll begin to use a test driven development approach and
>>>> we're considering GNU coreutils for our initial test case. I'd like to
>>>> know if it's possible to build only small parts of coreutils, instead
>>>> of the full package, during this initial evolution of libposix code.
>>>> Ideally, it would be good if we were able to build only a single tool
>>>> and its dependecies. Also, if that's not possible, maybe you could
>>>> point other applications that would be useful to us. I'm also afraid
>>>> that the configure script would not work with libposix in this stage.
>>>> Is it possible to point out what are the requirements for the
>>>> configure script ?
>>>
>>> The programs in coreutils depend heavily on gnulib.
>>> The parts of gnulib used in coreutils are distributed throughout the
>>> m4/*.m4 files (which are integrated into configure), lib/* files, and
>>> Makefile snippets.  The sources and gnulib support are intertwined
>>> enough that separating the src/*.c code from configure-time tests
>>> is likely to be a challenging and subtly error-prone task.
>>>
>>
>>
>>
>> --
>>  Henrique Dante de Almeida
>>  hd...@gm...
>>
>> ------------------------------------------------------------------------------
>> Crystal Reports - New Free Runtime and 30 Day Trial
>> Check out the new simplified licensing option that enables unlimited
>> royalty-free distribution of the report engine for externally facing
>> server and web deployment.
>> http://p.sf.net/sfu/businessobjects
>> _______________________________________________
>> Libposix-development mailing list
>> Lib...@li...
>> https://lists.sourceforge.net/lists/listinfo/libposix-development
>> ------------------------------------------------------------------------------
>> Crystal Reports - New Free Runtime and 30 Day Trial
>> Check out the new simplified licensing option that enables unlimited
>> royalty-free distribution of the report engine for externally facing
>> server and web deployment.
>> http://p.sf.net/sfu/businessobjects
>> _______________________________________________
>> Libposix-development mailing list
>> Lib...@li...
>> https://lists.sourceforge.net/lists/listinfo/libposix-development
>>
>
>
>
> --
>  Henrique Dante de Almeida
>  hd...@gm...
>



-- 
 Henrique Dante de Almeida
 hd...@gm...