time to release soon? GDB 8.1.1 release 2018-05-14 update

time to release soon? GDB 8.1.1 release 2018-05-14 update

[Joel Brobecker]

Hello everyone,

We're still down to one issue:

  * [KeithS/TomT] PR symtab/23010 (Regression: Assertion: SYMBOL_LANGUAGE (sym) == DICT_LANGUAGE (dict)->la_language)

       This is a regression. Tom sent a patch which got peer-reviewed
       by KeithS, and I gave it a nod, so it should go in sometime
       soon. But we still need to discuss whether the patch is suitable
       for 8.1.1.

       https://sourceware.org/ml/gdb-patches/2018-04/msg00234.html

We did get two PRs in the gdb-8.1-branch last week:

  * GDBserver broken on Windows
  * [AArch64] GDB cannot be used for debugging software that uses high Virtual Addresses

I propose we wait one or two more weeks for PR symtab/23010, decide
what want to do with this PR, and then release.

-- 
Joel

Re: time to release soon? GDB 8.1.1 release 2018-05-14 update

[Tom Tromey]

>>>>> "Joel" == Joel Brobecker <brobecker@adacore.com> writes:

Joel>        This is a regression. Tom sent a patch which got peer-reviewed
Joel>        by KeithS, and I gave it a nod, so it should go in sometime
Joel>        soon. But we still need to discuss whether the patch is suitable
Joel>        for 8.1.1.

Joel>        https://sourceware.org/ml/gdb-patches/2018-04/msg00234.html

Sorry, I didn't realize that the patch was approved.


Jan had also reported a crash with Rust and -readnow.
That requires the above patch, but also either:

https://sourceware.org/ml/gdb-patches/2018-04/msg00237.html

or

https://sourceware.org/ml/gdb-patches/2018-04/msg00353.html

... with the latter being preferred, though perhaps a bit invasive for a
backport.

Tom

Re: time to release soon? GDB 8.1.1 release 2018-05-14 update

[Joel Brobecker]

> >>>>> "Joel" == Joel Brobecker <brobecker@adacore.com> writes:
> 
> Joel>        This is a regression. Tom sent a patch which got peer-reviewed
> Joel>        by KeithS, and I gave it a nod, so it should go in sometime
> Joel>        soon. But we still need to discuss whether the patch is suitable
> Joel>        for 8.1.1.
> 
> Joel>        https://sourceware.org/ml/gdb-patches/2018-04/msg00234.html
> 
> Sorry, I didn't realize that the patch was approved.

Probably because I wasn't clear.

> Jan had also reported a crash with Rust and -readnow.
> That requires the above patch, but also either:
> 
> https://sourceware.org/ml/gdb-patches/2018-04/msg00237.html
> 
> or
> 
> https://sourceware.org/ml/gdb-patches/2018-04/msg00353.html
> 
> ... with the latter being preferred, though perhaps a bit invasive for a
> backport.

Agreed that the second patch is a little invasive. Looking at the first
one, it seems very localized, and can only affect Rust. If you feel
confident that the patch is safe, we could just create a PR number
for this issue, add it to the ChangeLog and revision log, and then
use version #1 for the 8.1 branch.

I started reviewing the tag-name-removal patch series, but ran out
of time. And now, it is looking like it's going to be a couple of weeks
before I can look at it again... Sorry!

-- 
Joel

Re: time to release soon? GDB 8.1.1 release 2018-05-14 update

[Tom Tromey]

>>>>> "Joel" == Joel Brobecker <brobecker@adacore.com> writes:

Joel> Agreed that the second patch is a little invasive. Looking at the first
Joel> one, it seems very localized, and can only affect Rust. If you feel
Joel> confident that the patch is safe, we could just create a PR number
Joel> for this issue, add it to the ChangeLog and revision log, and then
Joel> use version #1 for the 8.1 branch.

If the DWARF reader patch goes in 8.1, then I will do that.
Otherwise I probably won't.  I'm not very concerned about -readnow; I
think pretty much only gdb developers use it or should use it.

Tom

Re: time to release soon? GDB 8.1.1 release 2018-05-14 update

[Joel Brobecker]

> Joel> Agreed that the second patch is a little invasive. Looking at the first
> Joel> one, it seems very localized, and can only affect Rust. If you feel
> Joel> confident that the patch is safe, we could just create a PR number
> Joel> for this issue, add it to the ChangeLog and revision log, and then
> Joel> use version #1 for the 8.1 branch.
> 
> If the DWARF reader patch goes in 8.1, then I will do that.
> Otherwise I probably won't.  I'm not very concerned about -readnow; I
> think pretty much only gdb developers use it or should use it.

OK with me. But wouldn't the issue surface as well without -readnow,
when expanding the partial symtab?

Regarding the DWARF reader patch, I'd say we put it in 8.1.1.
We haven't had any strong feedback on that question, but the feedback
we received so far has been more in favor of including it.  I don't
think the patch is so risky that it should be excluded (most of it
is purely mechanical); and since Sergio mentioned Fedora was going
to apply it, we might as well give it to everyone ;-). Worse case
scenario, if it causes problems, I can always do an emergency 8.1.2.

-- 
Joel

Re: time to release soon? GDB 8.1.1 release 2018-05-14 update

[Tom Tromey]

>>>>> "Joel" == Joel Brobecker <brobecker@adacore.com> writes:

>> If the DWARF reader patch goes in 8.1, then I will do that.
>> Otherwise I probably won't.  I'm not very concerned about -readnow; I
>> think pretty much only gdb developers use it or should use it.

Joel> OK with me. But wouldn't the issue surface as well without -readnow,
Joel> when expanding the partial symtab?

In theory it shouldn't.

The way this works is that psymtab.c:psymtab_to_symtab, and other such
functions, should not try to expand a psymtab coming from a partial CU
(that both of these things are called "partial" is unfortunate).
Instead, these functions find the canonical user and expand that
instead, like:

  /* If it is a shared psymtab, find an unshared psymtab that includes
     it.  Any such psymtab will do.  */
  while (pst->user != NULL)
    pst = pst->user;


-readnow wasn't obeying this rule.  The fix isn't in psymtab.c, though,
because dwarf2read plays a little game to pretend that it is using the
index with -readnow, because in this mode there's no reason to make a
psymtab at all.

Maybe should still patch psym_expand_all_symtabs, but only DWARF is
using this feature, so there didn't seem to be a reason.

Joel> Regarding the DWARF reader patch, I'd say we put it in 8.1.1.

Ok, I will backport it next week.

Tom