Announce: Revisiting Patchwork (aka, if you have a patch in patchwork, your attention is ***required***!)

Dan Wilder Dan.Wilder@watchguard.com
Tue Dec 16 04:09:00 GMT 2014


Believe 4.9.2 needs it as well.   It's been repro'd there and the patched files are near-identical in the vicinity of the patch.   I'll look at knocking off a quick 4.9.2 to verify that.

The bug repros even on a platform built against much older gcc if the test program is static linked, so it's pretty easy to test.

--
Dan Wilder
________________________________________
From: Bryan Hundven [bryanhundven@gmail.com]
Sent: Monday, December 15, 2014 6:07 PM
To: Dan Wilder
Cc: crossgcc@sourceware.org
Subject: Re: Announce: Revisiting Patchwork (aka, if you have a patch in patchwork, your attention is ***required***!)

Dan,

On Mon, Dec 15, 2014 at 4:43 PM, Dan Wilder <Dan.Wilder@watchguard.com> wrote:
> OK those two patches under note 4, with a little backport work, did the trick for gcc-4.8.3.
>
> Any words of wisdom on submitting the patches to ctng and/or gcc?  I've got a git clone of crosstool-ng on github.

I say if you have a clone on github and can open a pull request, do that.

>
> My own repro and also the repro on gcc bug ran without error when built with -static using the patched compiler, on a newport.  Without the patch, boom.

Just curious, did you also test this with 4.9.2? IOW, does 4.9.x need
this patch as well?

> Running make check on the x86 build now.

Great!

Thanks,

-Bryan

> --
> Dan Wilder
> ________________________________________
> From: Bryan Hundven [bryanhundven@gmail.com]
> Sent: Thursday, December 04, 2014 6:01 PM
> To: Dan Wilder
> Cc: crossgcc@sourceware.org
> Subject: Re: Announce: Revisiting Patchwork (aka, if you have a patch in patchwork, your attention is ***required***!)
>
> Dan,
>
> On Thu, Dec 4, 2014 at 5:46 PM, Dan Wilder <Dan.Wilder@watchguard.com> wrote:
>> Thanks Bryan.  Yup their backtraces look just like mine.
>
> :)
>
>> I'll keep an eye on that bug, if somebody fixes it before I get a Round Tuit that'd be nice for me!  Otherwise I guess the patch goes upstream.
>
> Indeed. You can always add yourself to "cc" on the bug. I saw some
> other related bugs in my search, though they are "AIX"... that is
> still "powerpc" (rs6000).
>
> Also, Check the patches in comment #4.
>
> As a related... but slightly off-topic thought... Another thing I'd
> like to change about crosstool-ng: I dislike all these "for upstream"
> patches we harbor in patches/. Maybe I'll work on a RFC on how we can
> deal with them.
>
> Cheers,
>
> -Bryan
>
>> --
>> Dan Wilder
>> ________________________________________
>> From: Bryan Hundven [bryanhundven@gmail.com]
>> Sent: Thursday, December 04, 2014 5:42 PM
>> To: Dan Wilder
>> Cc: crossgcc@sourceware.org
>> Subject: Re: Announce: Revisiting Patchwork (aka, if you have a patch in patchwork, your attention is ***required***!)
>>
>> Dan,
>>
>> On Thu, Dec 4, 2014 at 4:24 PM, Dan Wilder <Dan.Wilder@watchguard.com> wrote:
>>> Any unwind patches for ppc in glibc-2.19?
>>
>> I don't see any in patchwork.
>>
>>> A small c++ program fails nicely on 32-bit ppc:
>>> ------------------------------------------------------------------------------------------
>>> #include <iostream>
>>>
>>> static void cause_exn(void)
>>> {
>>>     std::cout << "cause_exn invoked" << std::endl;
>>>     throw 20;
>>>     std::cout << "cause_exn done" << std::endl;
>>> }
>>>
>>> int main(int argc, char **argv)
>>> {
>>>     std::cout << "Starting" << std::endl;
>>>     try {
>>>         cause_exn();
>>>     } catch (int e) {
>>>         std::cout <<  "Caught exception " << e << std::endl;
>>>     }
>>>     std::cout << "Exiting" << std::endl;
>>>     return 0;
>>> }
>>> -----------------------------------------------------------------------------------------------
>>>
>>> Expected output is:
>>>
>>> Starting
>>> cause_exn invoked
>>> Caught exception 20
>>> Exiting
>>>
>>> on 32-bit ppc with glibc-2.19/gcc-4.8.3:
>>>
>>> Starting
>>> cause_exn invoked
>>> Aborted (core dumped)
>>>
>>> Have done some preliminary work and will have a patch for the list in a few weeks, but it would be a shame if somebody else has already gone there!
>>
>> Have you checked upstream?
>> https://sourceware.org/git/?p=glibc.git;a=shortlog;h=refs/heads/master
>>
>>> ppc_64 works fine.  As does x86, x86_64 on the same test program.
>>
>> Hrm...
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62231
>> Related?
>>
>>> --
>>> Dan Wilder
>>> Btw thanks Bryan for taking the cross-gcc helm and thanks Yann for the excellent work!
>>
>> :)
>>
>>>
>>>
>>> --
>>> For unsubscribe information see http://sourceware.org/lists.html#faq
>>>
>>
>> Cheers,
>>
>> -Bryan

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



More information about the crossgcc mailing list