GDB 7.9 release update

GDB 7.9 release update

[Joel Brobecker]

Hello,

Just a quick update on the 7.9 release process.

We have 10 entries listed in our TODO file.

| [Joel] PR gdb/17856 (ada-lang.c resaves in cache symbol it just found in cache)
| [Doug] PR gdb/17855 (clear_symtab_users doesn't call observer_notify_new_objfile (NULL) early enough)
| [Joel] PR gdb/17854 (Nothing sets pspace_data->sym_cache)

Patches were posted a while ago for those, so it's a matter of pushing
them. Doug, let's do it?

| [Doug] PR gdb/17821 (anonymous namespace + .gdb_index issue)

Status?

| [Doug] PR gdb/17799 (symbol lookup cache patch)

Defered for next release?

| [Doug] PR gdb/17798 (explicit linespecs patch)

I saw a recent message on this, so hopefully OK soon?

| [Doug] PR gdb/17681 (tab-completion needs better throttling)

I also saw the patch being discussed, so hopefully soon as well?

| [Phil?] PR cli/17716 ([7.7 regression] Backtrace cannot be stopped by 'q' or CTRL-C)

Phil?

| [Andreas Arnez] PR corefiles/17808 (7.9 regression: internal-error: i386_supply_gregset: Assertion `len == tdep->sizeof_gregset' failed.)

I think I saw a conditional OK on this patch.

| [Jan-pending] patch mail (compile: Filter out -fpreprocessed) (optional)

The patch needs to be reviewed.

Thanks!

-- 
Joel

Re: GDB 7.9 release update

[Doug Evans]

On Mon, Jan 26, 2015 at 8:37 PM, Joel Brobecker <brobecker@adacore.com> wrote:
> Hello,
>
> Just a quick update on the 7.9 release process.
>
> We have 10 entries listed in our TODO file.
>
> | [Joel] PR gdb/17856 (ada-lang.c resaves in cache symbol it just found in cache)
> | [Doug] PR gdb/17855 (clear_symtab_users doesn't call observer_notify_new_objfile (NULL) early enough)
> | [Joel] PR gdb/17854 (Nothing sets pspace_data->sym_cache)
>
> Patches were posted a while ago for those, so it's a matter of pushing
> them. Doug, let's do it?

Sure.

> | [Doug] PR gdb/17821 (anonymous namespace + .gdb_index issue)
>
> Status?
>
> | [Doug] PR gdb/17799 (symbol lookup cache patch)
>
> Defered for next release?
>
> | [Doug] PR gdb/17798 (explicit linespecs patch)
>
> I saw a recent message on this, so hopefully OK soon?
>
> | [Doug] PR gdb/17681 (tab-completion needs better throttling)
>
> I also saw the patch being discussed, so hopefully soon as well?

Hi.  I removed 17821, 17799, 17798 and 17681.
I'm just not going to get time to finish them right now.
They weren't super important though.

OTOH, I just found a bug that I'd like to make a blocker for 7.9.

I'm adding support for Two Level Linetables:
http://wiki.dwarfstd.org/index.php?title=TwoLevelLineTables

and when I run gdb on programs compiled with this support
gdb goes into an infinite loop. :-(

There are actually two bugs here:
17890: gdb needs to punt when it sees a dwarf line table header
version it doesn't understand
17891: how is SIGFPE handling supposed to work?  currently gdb goes
into an infinite loop

I don't have a strong opinion on getting 17891 fixed for 7.9,
but I would like 7.9 to not hang when given tll-enabled binaries.
The fix is safe and easy enough.

Re: GDB 7.9 release update

[Joel Brobecker]

> Hi.  I removed 17821, 17799, 17798 and 17681.
> I'm just not going to get time to finish them right now.
> They weren't super important though.

OK, thanks for doing that.

> OTOH, I just found a bug that I'd like to make a blocker for 7.9.
> 
> I'm adding support for Two Level Linetables:
> http://wiki.dwarfstd.org/index.php?title=TwoLevelLineTables

Sounds good.

I am also wondering if we should also wait for...
https://sourceware.org/bugzilla/show_bug.cgi?id=17837 ... or not.
Pedro, I see you've been kind enough to start looking into this
(bugzilla papertrail). What do you think? Blocker, not blocker?

Thanks you!
-- 
Joel

Re: GDB 7.9 release update

[Andreas Arnez]

On Tue, Jan 27 2015, Joel Brobecker wrote:

> Hello,
>
> Just a quick update on the 7.9 release process.
>
> We have 10 entries listed in our TODO file.
>
> ...
>
> | [Andreas Arnez] PR corefiles/17808 (7.9 regression: internal-error: i386_supply_gregset: Assertion `len == tdep->sizeof_gregset' failed.)
>
> I think I saw a conditional OK on this patch.

I've pinged it again:

  https://sourceware.org/ml/gdb-patches/2015-01/msg00755.html

Re: GDB 7.9 release update

[Pedro Alves]

On 01/29/2015 05:53 AM, Joel Brobecker wrote:

> I am also wondering if we should also wait for...
> https://sourceware.org/bugzilla/show_bug.cgi?id=17837 ... or not.
> Pedro, I see you've been kind enough to start looking into this
> (bugzilla papertrail). What do you think? Blocker, not blocker?

I'm not sure.  I haven't fully understood what the scripts in
question are doing and how the issue triggers.  I've asked Jan
if he could come up with a simplified reproducer.  Hopefully we'll
have a better idea soon.


There was also this:

  https://sourceware.org/ml/gdb-patches/2015-01/msg00703.html

The issue there is that with both remote and record targets, GDB
can get messed up after 'query'.  That's a regression compared
to 7.8.

I think I'll go push that series to master ASAP.  I'll wait for
feedback on the plan before pushing into 7.9 though.

Thanks,
Pedro Alves

Re: GDB 7.9 release update

[Eli Zaretskii]

> Date: Tue, 03 Feb 2015 15:24:40 +0100
> From: Pedro Alves <palves@redhat.com>
> CC: gdb-patches <gdb-patches@sourceware.org>,        Phil Muldoon <pmuldoon@redhat.com>, arnez@vnet.linux.ibm.com
> 
> On 01/29/2015 05:53 AM, Joel Brobecker wrote:
> 
> > I am also wondering if we should also wait for...
> > https://sourceware.org/bugzilla/show_bug.cgi?id=17837 ... or not.
> > Pedro, I see you've been kind enough to start looking into this
> > (bugzilla papertrail). What do you think? Blocker, not blocker?
> 
> I'm not sure.  I haven't fully understood what the scripts in
> question are doing and how the issue triggers.  I've asked Jan
> if he could come up with a simplified reproducer.  Hopefully we'll
> have a better idea soon.
> 
> 
> There was also this:
> 
>   https://sourceware.org/ml/gdb-patches/2015-01/msg00703.html
> 
> The issue there is that with both remote and record targets, GDB
> can get messed up after 'query'.  That's a regression compared
> to 7.8.
> 
> I think I'll go push that series to master ASAP.  I'll wait for
> feedback on the plan before pushing into 7.9 though.

Can we please solve the TUI slow display on MS-Windows?  Two patches
were proposed, can we please commit one of them?

TIA

Re: GDB 7.9 release update

[Joel Brobecker]

> > I am also wondering if we should also wait for...
> > https://sourceware.org/bugzilla/show_bug.cgi?id=17837 ... or not.
> > Pedro, I see you've been kind enough to start looking into this
> > (bugzilla papertrail). What do you think? Blocker, not blocker?
> 
> I'm not sure.  I haven't fully understood what the scripts in
> question are doing and how the issue triggers.  I've asked Jan
> if he could come up with a simplified reproducer.  Hopefully we'll
> have a better idea soon.

OK. Thanks a lot for looking into this for us, Pedro.
I've added this item to the list as a "maybe".

> There was also this:
> 
>   https://sourceware.org/ml/gdb-patches/2015-01/msg00703.html
> 
> The issue there is that with both remote and record targets, GDB
> can get messed up after 'query'.  That's a regression compared
> to 7.8.
> 
> I think I'll go push that series to master ASAP.  I'll wait for
> feedback on the plan before pushing into 7.9 though.

I've added this as a maybe as well. The patches are a little large,
but don't necessarily seem scary, and if the failure is bad enough...
I think you'll have the best perspective to make the call. We will
wait for feedback if we have to.

We still have a little extra time regardless, as I don't see much
activity on

    https://sourceware.org/bugzilla/show_bug.cgi?id=17716

I'm not too surprised. I think Phil said he was going to be traveling
too.

Thanks, Pedro!
-- 
Joel

Re: GDB 7.9 release update

[Phil Muldoon]

On 04/02/15 04:11, Joel Brobecker wrote:
>>> I am also wondering if we should also wait for...
>>> https://sourceware.org/bugzilla/show_bug.cgi?id=17837 ... or not.
>>> Pedro, I see you've been kind enough to start looking into this
>>> (bugzilla papertrail). What do you think? Blocker, not blocker?
>>
>> I'm not sure.  I haven't fully understood what the scripts in
>> question are doing and how the issue triggers.  I've asked Jan
>> if he could come up with a simplified reproducer.  Hopefully we'll
>> have a better idea soon.
>
> OK. Thanks a lot for looking into this for us, Pedro.
> I've added this item to the list as a "maybe".
>
>> There was also this:
>>
>>   https://sourceware.org/ml/gdb-patches/2015-01/msg00703.html
>>
>> The issue there is that with both remote and record targets, GDB
>> can get messed up after 'query'.  That's a regression compared
>> to 7.8.
>>
>> I think I'll go push that series to master ASAP.  I'll wait for
>> feedback on the plan before pushing into 7.9 though.
>
> I've added this as a maybe as well. The patches are a little large,
> but don't necessarily seem scary, and if the failure is bad enough...
> I think you'll have the best perspective to make the call. We will
> wait for feedback if we have to.
>
> We still have a little extra time regardless, as I don't see much
> activity on
>
>     https://sourceware.org/bugzilla/show_bug.cgi?id=17716
>
> I'm not too surprised. I think Phil said he was going to be traveling
> too.
>
> Thanks, Pedro!

Sorry for the delay. I've just come back from some meetings and then
FOSDEM. I'll get to work on this bug.  But again I do not consider it
a blocker, certainly not for release. There are workarounds for it
(disable frame-filter all), for the cases of exceptionally long
backtraces.

It is  unfortunately a non trivial issue having to do with both Python
interrupts and GDB exceptions, so a fix has to be carefully tested.

Cheers

Phil

Re: GDB 7.9 release update

[Joel Brobecker]

> Sorry for the delay. I've just come back from some meetings and then
> FOSDEM. I'll get to work on this bug.  But again I do not consider it
> a blocker, certainly not for release. There are workarounds for it
> (disable frame-filter all), for the cases of exceptionally long
> backtraces.
> 
> It is  unfortunately a non trivial issue having to do with both Python
> interrupts and GDB exceptions, so a fix has to be carefully tested.

Thanks for the update, Phil. Since this is non-trivial, and
therefore potentially risky for the branch, let's put that one
into the "Maybe" category.

Let's re-eval the state of the branch again on Wed (Feb 12th).

-- 
Joel

ready GDB 7.9 release? (was: "Re: GDB 7.9 release update")

[Joel Brobecker]

Hello everyone,

As far as I can tell, we are about ready to make the release.
There are 2 items on the TODO list for 7.9, but we have the patches
either approved or already checked in master.

So I will wait for a couple of days, and if there are no last-minute
issues, I will check the patches in, and create the 7.9 release.

Fingers crossedly yours,
-- 
Joel

Re: ready GDB 7.9 release?

[Pedro Alves]

On 02/17/2015 10:34 AM, Joel Brobecker wrote:
> Hello everyone,
> 
> As far as I can tell, we are about ready to make the release.
> There are 2 items on the TODO list for 7.9, but we have the patches
> either approved or already checked in master.
> 
> So I will wait for a couple of days, and if there are no last-minute
> issues, I will check the patches in, and create the 7.9 release.

Would it be OK to merge in the fixes for:

  [Pedro] (Fix racy FAILs of sigall-reverse.exp (and more))

      Now in master, gaining additional exposure. Need to decide whether to push to branch or not based on that.

to the branch?  The series has been on master since 3 Feb, and
I'm not aware of any related regression.  I'll do it today if OK.

(Although exposed by sigall-reverse.exp first, this affects any
"query" done while the target is installed in the event loop.)

Thanks,
Pedro Alves

Re: ready GDB 7.9 release?

[Joel Brobecker]

> Would it be OK to merge in the fixes for:
> 
>   [Pedro] (Fix racy FAILs of sigall-reverse.exp (and more))
> 
>       Now in master, gaining additional exposure. Need to decide whether to push to branch or not based on that.
> 
> to the branch?  The series has been on master since 3 Feb, and
> I'm not aware of any related regression.  I'll do it today if OK.
> 
> (Although exposed by sigall-reverse.exp first, this affects any
> "query" done while the target is installed in the event loop.)

Sounds good, please go ahead!

-- 
Joel

Re: ready GDB 7.9 release?

[Pedro Alves]

On 02/17/2015 11:21 AM, Joel Brobecker wrote:
>> Would it be OK to merge in the fixes for:
>>
>>   [Pedro] (Fix racy FAILs of sigall-reverse.exp (and more))
>>
>>       Now in master, gaining additional exposure. Need to decide whether to push to branch or not based on that.
>>
>> to the branch?  The series has been on master since 3 Feb, and
>> I'm not aware of any related regression.  I'll do it today if OK.
>>
>> (Although exposed by sigall-reverse.exp first, this affects any
>> "query" done while the target is installed in the event loop.)
> 
> Sounds good, please go ahead!

Done.  I've merged in 80bd5fab (Fix build breakage due to
event loop simplification) to the patch that introduced the
issue, regtested on x86_64 Fedora 20, and pushed.

Thanks,
Pedro Alves