This is the mail archive of the
gdb@sourceware.org
mailing list for the GDB project.
Re: [patch] Fix CLONE_VM vs. TLS [Re: Is CLONE_VM really needed in gdbserver?]
On Fri, Jan 29, 2010 at 02:26:25AM +0100, Jan Kratochvil wrote:
> CodeSourcery arm-2009q3-66-arm-uclinuxeabi-i686-pc-linux-gnu.tar.bz2 gcc sets
> __uClinux__ although gdb/gdbserver/linux-low.c depends on __UCLIBC__.
> CLONE_VM is a kernel feature so gcc should be more appropriate but maybe it
> does not matter.
It's the __UCLIBC_HAS_MMU__ / __ARCH_HAS_MMU__ that make this OK; the end
result is functionally equivalent to a __uClinux__ check. uClibc can
be used on full Linux too, and there you have fork.
>
>
> Thanks,
> Jan
>
>
> gdb/gdbserver/
> 2010-01-29 Jan Kratochvil <jan.kratochvil@redhat.com>
>
> PR libc/11214:
> * linux-low.c (linux_tracefork_child) [!(__UCLIBC__ && HAS_NOMMU)]: New.
> (linux_test_for_tracefork): Move `stack' into [__UCLIBC__ && HAS_NOMMU].
> (linux_test_for_tracefork) [!(__UCLIBC__ && HAS_NOMMU)]: New.
>
> gdb/testsuite/
> 2010-01-29 Jan Kratochvil <jan.kratochvil@redhat.com>
>
> PR libc/11214:
> * gdb.threads/current-lwp-dead.c (fn, main): Move CLONE_VM into
> [__uClinux__].
OK.
--
Daniel Jacobowitz
CodeSourcery