GDB 8.0 release / imminent branching? (2017-04-10)

GDB 8.0 release / imminent branching? (2017-04-10)

[Joel Brobecker]

Hello,

We're now about 4 weeks after our tentative branching date, and
I am concerned that if we continue to wait for some of the pending
issues to be resolved, we'll have new issues appearing on master
that'll might delay the branching further. So I'd like to propose
we branch in the next couple of days, and then backport fixes,
so as to stabilize the branch.

We have two 8.0 pending PRs:

* server/21169	GDBServer on ARM can crash the inferior while single stepping

     Antoine indicated that he's working on it, and his expected
     completion date isn't all that far away. Still on track for
     that, Antoine?

* build/21187	Static linking of libstdc++ and libgcc with GDB leads to broken exception handling on AIX platform

     For this one, I think the feedback so far is a question about
     the fact that this is only a work around. Parallel to that,
     did I see a thread where people are discussing a real fix
     for the exception handling issue?

     So, perhaps we might not need the GDB patch after all?
     Is there a version of GCC capable of compiling GDB 8.x
     on AiX?

Then, there is the issue of the Python APIs for record and btrace.
It sounds like we're getting close? Perhaps what we could do is
also backport the changes once the an agreement has been reached?

And finally, there is the build-bot. I'm having trouble using
the interface and understanding whether the current overall
state is rather positive or negative. Sergio, would you be able
to help me? If all good, I could create the branch with a SHA1
that you'd give me!

Thank you!
-- 
Joel

Re: GDB 8.0 release / imminent branching? (2017-04-10)

[Yao Qi]

Joel Brobecker <brobecker@adacore.com> writes:

> Then, there is the issue of the Python APIs for record and btrace.
> It sounds like we're getting close? Perhaps what we could do is
> also backport the changes once the an agreement has been reached?

The agreement on the interface has been reached and Tim is working on
it, as far as I know.  The Python API is a release blocker, but it
doesn't block branching.

-- 
Yao (齐尧)

Re: GDB 8.0 release / imminent branching? (2017-04-10)

[Andreas Arnez]

On Mon, Apr 10 2017, Joel Brobecker wrote:

> We have two 8.0 pending PRs:
>
> * server/21169	GDBServer on ARM can crash the inferior while single stepping
>
>      Antoine indicated that he's working on it, and his expected
>      completion date isn't all that far away. Still on track for
>      that, Antoine?
>
> * build/21187	Static linking of libstdc++ and libgcc with GDB leads to broken exception handling on AIX platform
>
>      For this one, I think the feedback so far is a question about
>      the fact that this is only a work around. Parallel to that,
>      did I see a thread where people are discussing a real fix
>      for the exception handling issue?
>
>      So, perhaps we might not need the GDB patch after all?
>      Is there a version of GCC capable of compiling GDB 8.x
>      on AiX?

Maybe not be relevant for answering your question, but the issue doesn't
seem to be restricted to AIX.  It occurs on Windows with mingw as well,
see gdb/21078 ("gdb compiled with mingw32 crashes on Windows7 when
entering any invalid command").

In addition to those PRs, It'd be great to get some of the DWARF piece
handling fixes in GDB 8.0, particularly the fix for gdb/21226.  There's
a patch set available that "just" needs review:

  https://sourceware.org/ml/gdb-patches/2017-04/msg00177.html

Also, I wonder who can look at exp/21162 ("GDB internal error with
unbounded array typecast expression").  It's a regression introduced
when adding variable length array support in April 2014.  The author,
Sanimir Agovic, doesn't seem to be active in GDB development any more?
The patch was reviewed by Joel Brobecker, though... ;-)

And last week a GDB internal error in value.c was reported as gdb/21356.
I haven't checked whether that is a regression or not.

--
Andreas

Re: GDB 8.0 release / imminent branching? (2017-04-10)

[Yao Qi]

Andreas Arnez <arnez@linux.vnet.ibm.com> writes:

> Maybe not be relevant for answering your question, but the issue doesn't
> seem to be restricted to AIX.  It occurs on Windows with mingw as well,
> see gdb/21078 ("gdb compiled with mingw32 crashes on Windows7 when
> entering any invalid command").

and Darwin host too,
https://sourceware.org/bugzilla/show_bug.cgi?id=21240

-- 
Yao (齐尧)

Re: GDB 8.0 release / imminent branching? (2017-04-10)

[Eli Zaretskii]

> From: Andreas Arnez <arnez@linux.vnet.ibm.com>
> Cc: gdb-patches@sourceware.org, Sergio Durigan Junior <sergiodj@redhat.com>,        antoine.tremblay@ericsson.com, mishra.nitish.88@gmail.com
> Date: Tue, 11 Apr 2017 16:21:27 +0200
> 
> On Mon, Apr 10 2017, Joel Brobecker wrote:
> 
> > * build/21187	Static linking of libstdc++ and libgcc with GDB leads to broken exception handling on AIX platform
> >
> >      For this one, I think the feedback so far is a question about
> >      the fact that this is only a work around. Parallel to that,
> >      did I see a thread where people are discussing a real fix
> >      for the exception handling issue?
> >
> >      So, perhaps we might not need the GDB patch after all?
> >      Is there a version of GCC capable of compiling GDB 8.x
> >      on AiX?
> 
> Maybe not be relevant for answering your question, but the issue doesn't
> seem to be restricted to AIX.  It occurs on Windows with mingw as well,
> see gdb/21078 ("gdb compiled with mingw32 crashes on Windows7 when
> entering any invalid command").

I cannot reproduce gdb/21078 in GDB 7.12 built with MinGW using static
linking of libstdc++ and libgcc.  I tried on Windows XP.  Maybe the OP
has some other issue with his build, like some inconsistency in how
exceptions are handled?

Re: GDB 8.0 release / imminent branching? (2017-04-10)

[Antoine Tremblay]

Joel Brobecker writes:

> Hello,
>
> We're now about 4 weeks after our tentative branching date, and
> I am concerned that if we continue to wait for some of the pending
> issues to be resolved, we'll have new issues appearing on master
> that'll might delay the branching further. So I'd like to propose
> we branch in the next couple of days, and then backport fixes,
> so as to stabilize the branch.
>
> We have two 8.0 pending PRs:
>
> * server/21169	GDBServer on ARM can crash the inferior while single stepping
>
>      Antoine indicated that he's working on it, and his expected
>      completion date isn't all that far away. Still on track for
>      that, Antoine?

Yes I just posted a patch with a partial fix for that.  I'm afraid the
full fix will have to wait however. 

Re: GDB 8.0 release / imminent branching? (2017-04-10)

[Sergio Durigan Junior]

On Monday, April 10 2017, Joel Brobecker wrote:

> Hello,
>
> We're now about 4 weeks after our tentative branching date, and
> I am concerned that if we continue to wait for some of the pending
> issues to be resolved, we'll have new issues appearing on master
> that'll might delay the branching further. So I'd like to propose
> we branch in the next couple of days, and then backport fixes,
> so as to stabilize the branch.

Yay :-).

[...]
> And finally, there is the build-bot. I'm having trouble using
> the interface and understanding whether the current overall
> state is rather positive or negative. Sergio, would you be able
> to help me? If all good, I could create the branch with a SHA1
> that you'd give me!

Sure thing.  I'll take a look this Wednesday and let you guys know if
there's anything suspicious there.

Cheers,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: GDB 8.0 release / imminent branching? (2017-04-10)

[Joel Brobecker]

Hi Andreas,

Thanks for pointing out more issues:

> Maybe not be relevant for answering your question, but the issue doesn't
> seem to be restricted to AIX.  It occurs on Windows with mingw as well,
> see gdb/21078 ("gdb compiled with mingw32 crashes on Windows7 when
> entering any invalid command").

Like Eli, I tried it, and it worked for me as well.

> In addition to those PRs, It'd be great to get some of the DWARF piece
> handling fixes in GDB 8.0, particularly the fix for gdb/21226.  There's
> a patch set available that "just" needs review:
> 
>   https://sourceware.org/ml/gdb-patches/2017-04/msg00177.html

If we weren't already 4 weeks late, I might have given ourselves
an extra week to see if it got accepted, but we need to branch
soon and start stabilizing. Maybe the Global Maintainer reviewing
the changes will approve the changes for a backport, but it looks
a bit on the large side. Since it's before a .0, perhaps it'll be
considered safe enough for it.

> Also, I wonder who can look at exp/21162 ("GDB internal error with
> unbounded array typecast expression").  It's a regression introduced
> when adding variable length array support in April 2014.  The author,
> Sanimir Agovic, doesn't seem to be active in GDB development any more?
> The patch was reviewed by Joel Brobecker, though... ;-)

I wish I could tell you that I would look into this, but my schedule
for the next few weeks is completely packed...

> And last week a GDB internal error in value.c was reported as gdb/21356.
> I haven't checked whether that is a regression or not.

I reproduced the issue with 7.10, so not a regression. I'd guess
that this is a missing check_typedef somewhere, so might be fairly
simple to fix. I've added this item to the "maybe" list for 8.0
in the release's wiki page
(https://sourceware.org/gdb/wiki/GDB_8.0_Release).

-- 
Joel

Re: GDB 8.0 release / imminent branching? (2017-04-10)

[Joel Brobecker]

> and Darwin host too,
> https://sourceware.org/bugzilla/show_bug.cgi?id=21240

FTR: I couldn't reproduce on Darwin either.

-- 
Joel