[PATCH] glibc: Refactor startfiles/headers into do_libc

Bryan Hundven bryanhundven@gmail.com
Thu Jun 23 02:42:00 GMT 2011


On Wed, Jun 22, 2011 at 3:50 PM, Yann E. MORIN
<yann.morin.1998@anciens.enib.fr> wrote:
> Bryan, All,
>
> On Monday 20 June 2011 09:44:21 Bryan Hundven wrote:
>> # HG changeset patch
>> # User Bryan Hundven <bryanhundven@gmail.com>
>> # Date 1308555696 25200
>> # Node ID b25856d9153785d0f711206383480f0190aec363
>> # Parent  2377be750e83defa6d764bcb8cf6dbaf098a19f6
>> glibc: Refactor startfiles/headers into do_libc
>
> As I said on IRC: I really like that idea!
>
>> The 'do_libc_start_files()' function was configuring libc differently then
>> 'do_libc()'. This might be fine for arm(eb) or powerpc(64), but not for
>> mips(64).
>>
>> Move the building of the libc start files and headers into do_libc, much like
>> how gcc handles stage1, stage2, and final toolchains.
>
> Did you check that:
>  - other archs were still OK?
>  - both glibc and eglibc were still OK?

I need to test more, but I was hoping by putting the patch out there,
I could get some more testing help?

   <_<    >_>

> Also, I'd prefer that we use a real backend function, a-la cc-core:
>  do_libc_start_files() {
>      do_libc_backend startfiles_mode=yes
>  }
>  do_libc() {
>      do_libc_backend startfiles_mode=no
>  }
>  do_libc_backend() {
>    # blabla
>  }

Cool, I like that too!

>> diff -r 2377be750e83 -r b25856d91537 scripts/build/libc/glibc-eglibc.sh-common
>> --- a/scripts/build/libc/glibc-eglibc.sh-common       Mon Jun 13 22:54:29 2011 +0400
>> +++ b/scripts/build/libc/glibc-eglibc.sh-common       Mon Jun 20 00:41:36 2011 -0700
> [--SNIP-]
>>  # This function builds and install the full C library
>>  do_libc() {
> [--SNIP-]
>> @@ -316,16 +218,75 @@
>>              ;;
>>      esac
>>
>> -    CT_DoLog EXTRA "Building C library"
>> -    CT_DoExecLog ALL make ${JOBSFLAGS}                      \
>> -                          "${extra_make_args[@]}"           \
>> -                          all
>> +    if [ "${startfile_mode}" = "yes" ]; then
>> +        CT_DoLog EXTRA "Installing C library headers"
>>
>> -    CT_DoLog EXTRA "Installing C library"
>> -    CT_DoExecLog ALL make ${JOBSFLAGS}                      \
>> -                          "${extra_make_args[@]}"           \
>> -                          install_root="${CT_SYSROOT_DIR}"  \
>> -                          install
>> +        # use the 'install-headers' makefile target to install the
>> +        # headers
>> +        CT_DoExecLog ALL make ${JOBSFLAGS}              \
>> +                         install_root=${CT_SYSROOT_DIR} \
>> +                         install-bootstrap-headers=yes  \
>> +                         install-headers
>
> What about the extra_make_args?
> It can contain preprocessor directives (eg. -DBROKEN_PPC_8xx_CPU15), so it
> might have an impact on headers...

Good point. I will add that to the new patch too.

>> +        # For glibc, a few headers need to be manually installed
>> +        if [ "${CT_LIBC}" = "glibc" ]; then
>> +            # Two headers -- stubs.h and features.h -- aren't installed by install-headers,
>> +            # so do them by hand.  We can tolerate an empty stubs.h for the moment.
>> +            # See e.g. http://gcc.gnu.org/ml/gcc/2002-01/msg00900.html
>> +            mkdir -p "${CT_HEADERS_DIR}/gnu"
>> +            CT_DoExecLog ALL touch "${CT_HEADERS_DIR}/gnu/stubs.h"
>> +            CT_DoExecLog ALL cp -v "${CT_SRC_DIR}/glibc-${CT_LIBC_VERSION}/include/features.h"  \
>> +                                   "${CT_HEADERS_DIR}/features.h"
>> +
>> +            # Building the bootstrap gcc requires either setting inhibit_libc, or
>> +            # having a copy of stdio_lim.h... see
>> +            # http://sources.redhat.com/ml/libc-alpha/2003-11/msg00045.html
>> +            CT_DoExecLog ALL cp -v bits/stdio_lim.h "${CT_HEADERS_DIR}/bits/stdio_lim.h"
>> +
>> +            # Following error building gcc-4.0.0's gcj:
>> +            #  error: bits/syscall.h: No such file or directory
>> +            # solved by following copy; see http://sourceware.org/ml/crossgcc/2005-05/msg00168.html
>> +            # but it breaks arm, see http://sourceware.org/ml/crossgcc/2006-01/msg00091.html
>> +            case "${CT_ARCH}" in
>> +                arm)    ;;
>> +                *)  CT_DoExecLog ALL cp -v "misc/syscall-list.h"            \
>> +                                           "${CT_HEADERS_DIR}/bits/syscall.h"
>> +                    ;;
>> +            esac
>> +        fi
>> +
>> +        if [ "${CT_THREADS}" = "nptl" ]; then
>> +            CT_DoLog EXTRA "Installing C library start files"
>> +
>> +            # there are a few object files needed to link shared libraries,
>> +            # which we build and install by hand
>> +            CT_DoExecLog ALL mkdir -p "${CT_SYSROOT_DIR}/usr/lib"
>> +            CT_DoExecLog ALL make ${JOBSFLAGS} csu/subdir_lib
>
> Ditto extra_make_args.

ack.

>> +            CT_DoExecLog ALL cp csu/crt1.o csu/crti.o csu/crtn.o \
>> +                                "${CT_SYSROOT_DIR}/usr/lib"
>> +
>> +            # Finally, 'libgcc_s.so' requires a 'libc.so' to link against.
>> +            # However, since we will never actually execute its code,
>> +            # it doesn't matter what it contains.  So, treating '/dev/null'
>> +            # as a C source file, we produce a dummy 'libc.so' in one step
>> +            CT_DoExecLog ALL "${cross_cc}" -nostdlib        \
>> +                                           -nostartfiles    \
>> +                                           -shared          \
>> +                                           -x c /dev/null   \
>> +                                           -o "${CT_SYSROOT_DIR}/usr/lib/libc.so"
>> +        fi # threads == nptl
>> +    else # startfile_mode = no
>> +        CT_DoLog EXTRA "Building C library"
>> +        CT_DoExecLog ALL make ${JOBSFLAGS}                      \
>> +                              "${extra_make_args[@]}"           \
>> +                              all
>> +
>> +        CT_DoLog EXTRA "Installing C library"
>> +        CT_DoExecLog ALL make ${JOBSFLAGS}                      \
>> +                              "${extra_make_args[@]}"           \
>> +                              install_root="${CT_SYSROOT_DIR}"  \
>> +                              install
>> +    fi
>>
>>      CT_EndStep
>>  }
>> @@ -384,3 +345,5 @@
>>              ;;
>>      esac
>>  }
>> +
>> +# vim: ts=4 sw=4 et ai syn=sh
>
> Spurious line. We do not have vim syntax settings in the source code.

We could provide the combined emacs/vim syntax settings.
I like the idea of the editor "doing the right thing", and me just
writing code and not worrying about style.

> Although I do indeed exclusively use vim. ;-)

I use a lot of editors, and most of them either use vim or emacs style
'mode lines'.
I prefer the modeline in the code, as I have other projects where they like:
  ts=2 sw=2 et

and changing my .vimrc or .emacs for each project is a serious pain.
Also, opening a .in file, in this specific case, shows up as a plain-text file.
So my syntax highlighting goes away, and I have to bang on my space
bar... a lot!

It would be nice... my keyboard would thank you... but not totally necessary.

>
> Regards.
> Yann E. MORIN.
>
> --
> .-----------------.--------------------.------------------.--------------------.
> |  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
> | +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
> | +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
> | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
> '------------------------------^-------^------------------^--------------------'
>

-Bryan

--
For unsubscribe information see http://sourceware.org/lists.html#faq



More information about the crossgcc mailing list