This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Fix for Tilera TILE-Gx SIGTRAMP_FRAME unwinding
On 07/20/2012 06:41 PM, Jeff Kenton wrote:
> On 07/20/2012 01:24 PM, Pedro Alves wrote:
>> On 07/06/2012 07:59 PM, Jeff Kenton wrote:
>>> This is a fix for SIGTRAMP_FRAME unwinding for TILE-Gx. It now calculates offsets into the stack frame to properly locate saved registers. This eliminates all remaining testsuite failures from the sigstep test.
>> Thanks.
>>
>>> gdb/
>>> * tilegx-linux-tdep.c (tilegx_linux_sigframe_init): Fix SIGTRAMP_FRAME unwinding.
>> Looks fine from a mechanics perspective. I trust you on the TILE-Gx specifics, of course. :-)
>>
>>> static void
>>> tilegx_linux_init_abi (struct gdbarch_info info, struct gdbarch *gdbarch)
>>> {
>>> - int arch_size = gdbarch_addr_bit (gdbarch);
>>> + int arch_size;
>>>
>>> linux_init_abi (info, gdbarch);
>>>
>>> @@ -117,6 +127,7 @@
>>> tilegx_regset_from_core_section);
>>>
>>> /* GNU/Linux uses SVR4-style shared libraries. */
>>> + arch_size = gdbarch_addr_bit (gdbarch);
>>> if (arch_size == 32)
>>> set_solib_svr4_fetch_link_map_offsets (gdbarch,
>>> svr4_ilp32_fetch_link_map_offsets);
>> This isn't mentioned in the ChangeLog entry, and looks unrelated though?
>>
>
> Sorry. Unrelated. Trivial rearrangement of code. You can keep it or ignore it with no effect.
Okay, it was easier to drop it. I've applied the patch without that hunk. Send it along (in a
new thread) with a ChangeLog entry if you want it in.
--
Pedro Alves