[OE-core] [PATCH] gdb: Upgrade from 8.2.1 to 8.3

Khem Raj raj.khem at gmail.com
Sun May 19 05:08:15 UTC 2019



On 5/14/19 11:24 AM, Randy MacLeod wrote:
> On 5/14/19 1:36 PM, Richard Purdie wrote:
>> On Mon, 2019-05-13 at 22:53 +0000, Alistair Francis wrote:
>>> Signed-off-by: Alistair Francis <alistair.francis at wdc.com>
>>> ---
>>> If there are issues with the patch the branch is avaliable here:
>>> https://github.com/alistair23/openembedded-core/tree/alistair/gdb-8.3
>>
>> Seems there is a problem in oe-selftest:
>>
>> https://autobuilder.yoctoproject.org/typhoon/#/builders/86/builds/174/steps/7/logs/step2d 
>>
>>
>> "oe-selftest -r package.PackageTests.test_gdb_hardlink_debug" should
>> reproduce and it seemed to happen on more than one host.
> 
> I have some off-list notes, show below, for a bug that I haven't
> gotten to yet.
> 
> Alistair,
> 
> Could you adjust the oeqa match criteria based on the info below?
> 
> ../Randy
> 
> 
> The oe-selftest:
>     https://bugzilla.yoctoproject.org/show_bug.cgi?id=13123
> is failing because the check at:
>     meta/lib/oeqa/selftest/cases/package.py:
>                 elif "Breakpoint 1, main () at hello.c:4" in l:
> is too precise. (This wasn't obvious to me for quite a
> while based on the defect report!)
> 
> Anyway, at times, the test would pass for me other times
> it would not as you may have noticed. For interactive tests, gdb
> always functions properly but the breakpoint doesn't always
> result in the *exact* string:
>     "Breakpoint 1, main () at hello.c:4"
> sometimes (with glibc-dev, iirc) it's:
>     "Breakpoint 1, 0x0804905e in main () at hello.c:4"
> and sometimes it's just:
>     "Breakpoint 1, 0x0804905e: file hello.c, line 4"
> 

perhaps its better to check output of

i b 1

instead of relying on output of setting the breakpoint.

> so unless Khem or someone knows how to ensure
> 100% consistent behaviour, perhaps we should search for:
> 
>     "Breakpoint 1, .* hello.c.*4"   ( I'm not sure of the proper regex.)
> since that will confirm that a breakpoint was hit at the proper file/line.
> 
>>
>> Cheers,
>>
>> Richard
>>
> 
> 


More information about the Openembedded-core mailing list