GDB 7.8 update (getting closer, new pre-release)

GDB 7.8 update (getting closer, new pre-release)

[Joel Brobecker]

Hello everyone,

First, a big Thank You to everyone who worked on cleaning out
all the regressions and blocking issues so far. While the release
process for this branch started as the quietest ever for me,
the number of identified blocking issues eventually rose to about
3 times the number of issues identified in the previous release
process. We're down to 3 entries in the TODO list before 7.8 release,
so Thank You to everyone involved in the fixes getting us there.

The 3 entries are:

    [Doug/Pedro] Asynchronous + all-stop problematic on gdbserver
    (being debated as not critical for 7.8, and I think I saw a patch
    from Doug to help the user in case of problem)

    [Doug] /usr/share/gdb/{guile,python}/ gets installed even --without-{guile,python}
    NTH - not serious, nice to have only

    [Doug] Precompile scheme files
    NTH - not critical, but just a bit more important than "nice to have"

Issue #1 looks like it's on its way to being mitigated, and in
the meantime isn't critical. Issue #2 is "nice to have", and
issue #3 is "important".

Doug - any thoughts on when we might a fix, at least for issue #3?
I propose that we take issue #2 out of the mix. We've been doing
this for Python files too, so it could be argued as not being
a regression.

I propose I make a second pre-release tomorrow morning, or even
tonight. That should give people some time to re-test. I'm away
next week, but hopefully by the time I come back, we'll be ready
to make that release.

Thoughts?

-- 
Joel

Re: GDB 7.8 update (getting closer, new pre-release)

[H.J. Lu]

On Thu, Jul 17, 2014 at 6:38 AM, Joel Brobecker <brobecker@adacore.com> wrote:
> Hello everyone,
>
> First, a big Thank You to everyone who worked on cleaning out
> all the regressions and blocking issues so far. While the release
> process for this branch started as the quietest ever for me,
> the number of identified blocking issues eventually rose to about
> 3 times the number of issues identified in the previous release
> process. We're down to 3 entries in the TODO list before 7.8 release,
> so Thank You to everyone involved in the fixes getting us there.
>
> The 3 entries are:
>
>     [Doug/Pedro] Asynchronous + all-stop problematic on gdbserver
>     (being debated as not critical for 7.8, and I think I saw a patch
>     from Doug to help the user in case of problem)
>
>     [Doug] /usr/share/gdb/{guile,python}/ gets installed even --without-{guile,python}
>     NTH - not serious, nice to have only
>
>     [Doug] Precompile scheme files
>     NTH - not critical, but just a bit more important than "nice to have"
>
> Issue #1 looks like it's on its way to being mitigated, and in
> the meantime isn't critical. Issue #2 is "nice to have", and
> issue #3 is "important".
>
> Doug - any thoughts on when we might a fix, at least for issue #3?
> I propose that we take issue #2 out of the mix. We've been doing
> this for Python files too, so it could be argued as not being
> a regression.
>
> I propose I make a second pre-release tomorrow morning, or even
> tonight. That should give people some time to re-test. I'm away
> next week, but hopefully by the time I come back, we'll be ready
> to make that release.
>
> Thoughts?
>

Is that possible to include those 2 patches:

https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=144bed8d4d8a1bdc0067f55f2ee71c07e5594677
https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=cca5b8b64b3286bb96cd7a2c18232d1acea85cd9

in GDB 7.8. It properly disassembles PLT section in libc.so with
IFUNC.

Thanks.

-- 
H.J.

Re: GDB 7.8 update (getting closer, new pre-release)

[Doug Evans]

On Thu, Jul 17, 2014 at 2:38 PM, Joel Brobecker <brobecker@adacore.com> wrote:
> Hello everyone,
>
> First, a big Thank You to everyone who worked on cleaning out
> all the regressions and blocking issues so far. While the release
> process for this branch started as the quietest ever for me,
> the number of identified blocking issues eventually rose to about
> 3 times the number of issues identified in the previous release
> process. We're down to 3 entries in the TODO list before 7.8 release,
> so Thank You to everyone involved in the fixes getting us there.
>
> The 3 entries are:
>
>     [Doug/Pedro] Asynchronous + all-stop problematic on gdbserver
>     (being debated as not critical for 7.8, and I think I saw a patch
>     from Doug to help the user in case of problem)
>
>     [Doug] /usr/share/gdb/{guile,python}/ gets installed even --without-{guile,python}
>     NTH - not serious, nice to have only
>
>     [Doug] Precompile scheme files
>     NTH - not critical, but just a bit more important than "nice to have"
>
> Issue #1 looks like it's on its way to being mitigated, and in
> the meantime isn't critical.

Right, though I'd really like the mitigation to go in 7.8.
It's all set to go, just need to check in.

> Issue #2 is "nice to have", and
> issue #3 is "important".
>
> Doug - any thoughts on when we might a fix, at least for issue #3?

Working on it now. We have a patch, just need to finish it up.

> I propose that we take issue #2 out of the mix. We've been doing
> this for Python files too, so it could be argued as not being
> a regression.

Agreed.

> I propose I make a second pre-release tomorrow morning, or even
> tonight. That should give people some time to re-test. I'm away
> next week, but hopefully by the time I come back, we'll be ready
> to make that release.
>
> Thoughts?

ok by me.

Re: GDB 7.8 update (getting closer, new pre-release)

[Joel Brobecker]

> Is that possible to include those 2 patches:
> 
> https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=144bed8d4d8a1bdc0067f55f2ee71c07e5594677
> https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=cca5b8b64b3286bb96cd7a2c18232d1acea85cd9
> 
> in GDB 7.8. It properly disassembles PLT section in libc.so with
> IFUNC.

Can you please have one of the binutils maintainers assess the risk
level of backporting these patches into 7.8?

-- 
Joel

Re: GDB 7.8 update (getting closer, new pre-release)

[Doug Evans]

On Thu, Jul 17, 2014 at 7:13 PM, Doug Evans <dje@google.com> wrote:
>> I propose that we take issue #2 out of the mix. We've been doing
>> this for Python files too, so it could be argued as not being
>> a regression.
>
> Agreed.

Hi.
fyi, I need to do something in this area for compiled guile files, so
I'm going to tackle this for 7.8.

Re: Fwd: GDB 7.8 update (getting closer, new pre-release)

[Alan Modra]

> > https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=144bed8d4d8a1bdc0067f55f2ee71c07e5594677
> > https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=cca5b8b64b3286bb96cd7a2c18232d1acea85cd9
> >
> > in GDB 7.8. It properly disassembles PLT section in libc.so with
> > IFUNC.
> 
> Can you please have one of the binutils maintainers assess the risk
> level of backporting these patches into 7.8?

I've become a little rusty on the x86 target code but it looks OK to
me.  The simple approach that HJ has taken is quite suited to a
release branch, as compared to redesigning x86_64 ifunc support..
(The latter might be preferable on mainline to avoid the reloc
scanning, if possible, but I leave that decision to HJ.)

-- 
Alan Modra
Australia Development Lab, IBM

Re: Fwd: GDB 7.8 update (getting closer, new pre-release)

[Joel Brobecker]

> > > https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=144bed8d4d8a1bdc0067f55f2ee71c07e5594677
> > > https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=cca5b8b64b3286bb96cd7a2c18232d1acea85cd9
> > >
> > > in GDB 7.8. It properly disassembles PLT section in libc.so with
> > > IFUNC.
> > 
> > Can you please have one of the binutils maintainers assess the risk
> > level of backporting these patches into 7.8?
> 
> I've become a little rusty on the x86 target code but it looks OK to
> me.  The simple approach that HJ has taken is quite suited to a
> release branch, as compared to redesigning x86_64 ifunc support..
> (The latter might be preferable on mainline to avoid the reloc
> scanning, if possible, but I leave that decision to HJ.)

Thanks, Alan.

HJ,

if this hasn't been ported to the branch yet, please go ahead now.

-- 
Joel

Re: Fwd: GDB 7.8 update (getting closer, new pre-release)

[H.J. Lu]

On Mon, Jul 28, 2014 at 8:14 AM, Joel Brobecker <brobecker@adacore.com> wrote:
>> > > https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=144bed8d4d8a1bdc0067f55f2ee71c07e5594677
>> > > https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=cca5b8b64b3286bb96cd7a2c18232d1acea85cd9
>> > >
>> > > in GDB 7.8. It properly disassembles PLT section in libc.so with
>> > > IFUNC.
>> >
>> > Can you please have one of the binutils maintainers assess the risk
>> > level of backporting these patches into 7.8?
>>
>> I've become a little rusty on the x86 target code but it looks OK to
>> me.  The simple approach that HJ has taken is quite suited to a
>> release branch, as compared to redesigning x86_64 ifunc support..
>> (The latter might be preferable on mainline to avoid the reloc
>> scanning, if possible, but I leave that decision to HJ.)
>
> Thanks, Alan.
>
> HJ,
>
> if this hasn't been ported to the branch yet, please go ahead now.
>

I checked them into 7.8 branch.

Thanks.

-- 
H.J.

Re: GDB 7.8 update (getting closer, new pre-release)

[Edjunior Barbosa Machado]

On 07/17/2014 10:38 AM, Joel Brobecker wrote:
> Hello everyone,
> 
> First, a big Thank You to everyone who worked on cleaning out
> all the regressions and blocking issues so far. While the release
> process for this branch started as the quietest ever for me,
> the number of identified blocking issues eventually rose to about
> 3 times the number of issues identified in the previous release
> process. We're down to 3 entries in the TODO list before 7.8 release,
> so Thank You to everyone involved in the fixes getting us there.
> 
> The 3 entries are:
> 
>     [Doug/Pedro] Asynchronous + all-stop problematic on gdbserver
>     (being debated as not critical for 7.8, and I think I saw a patch
>     from Doug to help the user in case of problem)
> 
>     [Doug] /usr/share/gdb/{guile,python}/ gets installed even --without-{guile,python}
>     NTH - not serious, nice to have only
> 
>     [Doug] Precompile scheme files
>     NTH - not critical, but just a bit more important than "nice to have"
> 
> Issue #1 looks like it's on its way to being mitigated, and in
> the meantime isn't critical. Issue #2 is "nice to have", and
> issue #3 is "important".
> 
> Doug - any thoughts on when we might a fix, at least for issue #3?
> I propose that we take issue #2 out of the mix. We've been doing
> this for Python files too, so it could be argued as not being
> a regression.
> 
> I propose I make a second pre-release tomorrow morning, or even
> tonight. That should give people some time to re-test. I'm away
> next week, but hopefully by the time I come back, we'll be ready
> to make that release.
> 
> Thoughts?

Hi,
is possible to check these 2 patches into 7.8 branch?

https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=aebf9d247e4026189a0f612ee03c19cdf4bf86ac
https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=572f65559b461133292118cee522b7e437714a07

These patches solve several unexpected failures from
gdb.base/watchpoint-reuse-slot.exp in ppc64 servers and s390/s390x due to
inappropriate handling on targets that don't support hw break/watchpoints.

Thanks,
--
Edjunior

Re: GDB 7.8 update (getting closer, new pre-release)

[Joel Brobecker]

> Hi,
> is possible to check these 2 patches into 7.8 branch?

Just in the nick of time! :-)

> https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=aebf9d247e4026189a0f612ee03c19cdf4bf86ac
> https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=572f65559b461133292118cee522b7e437714a07
> 
> These patches solve several unexpected failures from
> gdb.base/watchpoint-reuse-slot.exp in ppc64 servers and s390/s390x due to
> inappropriate handling on targets that don't support hw break/watchpoints.

Yes, those two are OK for 7.8.

-- 
Joel

Re: GDB 7.8 update (getting closer, new pre-release)

[Edjunior Barbosa Machado]

On 07/28/2014 05:20 PM, Joel Brobecker wrote:
>> Hi,
>> is possible to check these 2 patches into 7.8 branch?
> 
> Just in the nick of time! :-)
> 
>> https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=aebf9d247e4026189a0f612ee03c19cdf4bf86ac
>> https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=572f65559b461133292118cee522b7e437714a07
>>
>> These patches solve several unexpected failures from
>> gdb.base/watchpoint-reuse-slot.exp in ppc64 servers and s390/s390x due to
>> inappropriate handling on targets that don't support hw break/watchpoints.
> 
> Yes, those two are OK for 7.8.

Thanks Joel, I've just checked them into 7.8 branch:

	https://sourceware.org/ml/gdb-cvs/2014-07/msg00100.html

--
Edjunior