Re: [PULL 00/30] testing, gdbstub and semihosting

[Alex Bennée]

Philippe Mathieu-Daudé <philmd@redhat.com> writes:

> Hi Alex,
>
> On 1/18/21 1:18 PM, Alex Bennée wrote:
>> 
>> Peter Maydell <peter.maydell@linaro.org> writes:
>> 
>>> On Fri, 15 Jan 2021 at 13:08, Alex Bennée <alex.bennee@linaro.org> wrote:
>>>>
>>>> The following changes since commit 
>>>> 7c79721606be11b5bc556449e5bcbc331ef6867d:
>>>>
>>>>   Merge remote-tracking branch 'remotes/rth-gitlab/tags/pull-tcg-20210113' 
>>>> into staging (2021-01-14 09:54:29 +0000)
>>>>
>>>> are available in the Git repository at:
>>>>
>>>>   https://github.com/stsquad/qemu.git tags/pull-testing-and-misc-150121-1
>>>>
>>>> for you to fetch changes up to be846761ca8b5a7e2e7b7108c8c156126b799824:
>>>>
>>>>   semihosting: Implement SYS_ISERROR (2021-01-15 11:12:34 +0000)
>>>>
>>>> ----------------------------------------------------------------
>>>> Testing, gdbstub and semihosting patches:
>>>>
>>>>   - clean-ups to docker images
>>>>   - drop duplicate jobs from shippable
>>>>   - prettier tag generation (+gtags)
>>>>   - generate browsable source tree
>>>>   - more Travis->GitLab migrations
>>>>   - fix checkpatch to deal with commits
>>>>   - gate gdbstub tests on 8.3.1, expand tests
>>>>   - support Xfer:auxv:read gdb packet
>>>>   - better gdbstub cleanup
>>>>   - use GDB's SVE register layout
>>>>   - make arm-compat-semihosting common
>>>>   - add riscv semihosting support
>>>>   - add HEAPINFO, ELAPSED, TICKFREQ, TMPNAM and ISERROR to semihosting
>>>
>>> Fails to build, netbsd:
>>>
>>> ../src/gdbstub.c: In function 'handle_query_xfer_auxv':
>>> ../src/gdbstub.c:2258:26: error: 'struct image_info' has no member
>>> named 'saved_auxv'
>>>      saved_auxv = ts->info->saved_auxv;
>>>                           ^~
>>> ../src/gdbstub.c:2259:24: error: 'struct image_info' has no member
>>> named 'auxv_len'
>>>      auxv_len = ts->info->auxv_len;
>> 
>> I've:
>> 
>> #if defined(CONFIG_USER_ONLY) && defined(CONFIG_LINUX_USER)
>> 
>> around the code so it won't build for the *BSDs.
>
> CONFIG_LINUX_USER implies CONFIG_USER_ONLY, right?

Probably could in a clean-up patch.

>
> Maybe long-term this can become:
>
> #if defined(CONFIG_LINUX_USER)
> #elif defined(...BSD...)
> #endif

Well the failure is missing the data in our own structures. I don't know
if BSD does support auxv, I guess if it's a POSIX thing. I guess we'll
see when we get the BSD rewrite.

> (maybe worth to fix if the pullreq isn't processed,
> else not a big deal).

Already merged so ¯\_(ツ)_/¯...

-- 
Alex Bennée