status on gdb-7.4 release...

status on gdb-7.4 release...

[Joel Brobecker]

... planned January 2nd.

Hello everyone,

We had the few usual late discoveries and hicups, but the 7.4 release
seems well under way. As far as I know, there are a couple of
regressions that have been detected and still not fixed:

 1. Regression gdb.base/fixsection.exp && gdb.threads/thread_check.exp
    http://www.sourceware.org/ml/gdb-patches/2011-12/msg00560.html

 2. 7.3->7.4 s390x regressions
    http://www.sourceware.org/ml/gdb/2011-12/msg00039.html

    I think that these ones were marked as being a latent bug
    caused by the kernel. Correct?
    http://www.sourceware.org/ml/gdb-patches/2011-12/msg00743.html

Anything else I might have missed?

Thanks,
-- 
Joel

Re: status on gdb-7.4 release...

[Ralf Corsepius]

On 12/22/2011 07:26 AM, Joel Brobecker wrote:
> ... planned January 2nd.
>
> Hello everyone,
>
> We had the few usual late discoveries and hicups, but the 7.4 release
> seems well under way. As far as I know, there are a couple of
> regressions that have been detected and still not fixed:
>
>   1. Regression gdb.base/fixsection.exp&&  gdb.threads/thread_check.exp
>      http://www.sourceware.org/ml/gdb-patches/2011-12/msg00560.html
>
>   2. 7.3->7.4 s390x regressions
>      http://www.sourceware.org/ml/gdb/2011-12/msg00039.html
>
>      I think that these ones were marked as being a latent bug
>      caused by the kernel. Correct?
>      http://www.sourceware.org/ml/gdb-patches/2011-12/msg00743.html
>
> Anything else I might have missed?

A yet unreported observation on my part:

gdb-7.3.91 installs $(includedir)/gdb/jit-reader.h

No idea what this is, why gdb does so and what it would be useful for.

To me it looks like a header without a corresponding library.

Ralf

Re: status on gdb-7.4 release...

[Joel Brobecker]

> gdb-7.3.91 installs $(includedir)/gdb/jit-reader.h
> No idea what this is, why gdb does so and what it would be useful for.
> To me it looks like a header without a corresponding library.

Thanks.

I don't know all the details, but this is the API that programs
should use if they want to provide their own reader for JIT
debug information. There is no need for a corresponding library
as far as I know. See:
http://www.sourceware.org/gdb/current/onlinedocs/gdb/JIT-Interface.html#JIT-Interface

-- 
Joel

Re: status on gdb-7.4 release...

[Ralf Corsepius]

On 12/22/2011 08:25 AM, Joel Brobecker wrote:
>> gdb-7.3.91 installs $(includedir)/gdb/jit-reader.h
>> No idea what this is, why gdb does so and what it would be useful for.
>> To me it looks like a header without a corresponding library.
> Thanks.
>
> I don't know all the details, but this is the API that programs
> should use if they want to provide their own reader for JIT
> debug information. There is no need for a corresponding library
> as far as I know. See:
> http://www.sourceware.org/gdb/current/onlinedocs/gdb/JIT-Interface.html#JIT-Interface
>
OK, then ... wouldn't it be better to install this header into a 
versioned directory, say

$(includedir)/gdb-$(GDB_VERSION)
or
$(includedir)/gdb/$(GDB_VERSION)
?

Otherwise this file will conflict with other gdbs' versions of this file when having different versions of gdb installed in parallel[1] and should this file change in future versions of gdb.

Ralf

[1] I am building cross-gdbs to be installed in parallel to a system's native gdb and other cross-gdbs.

Re: status on gdb-7.4 release...

[Joel Brobecker]

> OK, then ... wouldn't it be better to install this header into a
> versioned directory, say
> 
> $(includedir)/gdb-$(GDB_VERSION)
> or
> $(includedir)/gdb/$(GDB_VERSION)
> ?
> 
> Otherwise this file will conflict with other gdbs' versions of this
> file when having different versions of gdb installed in parallel[1]
> and should this file change in future versions of gdb.

The same is true of everything else that goes into the so-called
gdb-datadir, I think. What we should probably do is make the default
gdb-datadir versioned.  This is in fact what we manually do at AdaCore,
by configuring with --with-gdb-datadir.

But if this is something you'd like to pursue, it'd be better if
you did so in a separate thread. I'd like to keep this purely to
track the release. And the suggestion above have drawbacks, so
it's not a no-brainer.

-- 
Joel

Re: status on gdb-7.4 release...

[Ralf Corsepius]

On 12/22/2011 09:11 AM, Joel Brobecker wrote:
>> OK, then ... wouldn't it be better to install this header into a
>> versioned directory, say
>>
>> $(includedir)/gdb-$(GDB_VERSION)
>> or
>> $(includedir)/gdb/$(GDB_VERSION)
>> ?
>>
>> Otherwise this file will conflict with other gdbs' versions of this
>> file when having different versions of gdb installed in parallel[1]
>> and should this file change in future versions of gdb.
> The same is true of everything else that goes into the so-called
> gdb-datadir, I think.
Agreed.

> What we should probably do is make the default
> gdb-datadir versioned.  This is in fact what we manually do at AdaCore,
> by configuring with --with-gdb-datadir.
I so far have been using --with-gdb-datadir=$(datadir)/$(target)-gdb
Likely you are right and I should start using a versioned directory.

> But if this is something you'd like to pursue, it'd be better if
> you did so in a separate thread. I'd like to keep this purely to
> track the release. And the suggestion above have drawbacks, so
> it's not a no-brainer.
>
Agreed. I'll try to use a versioned --includedir and --with-gdb-datadir 
in my next build spin.

Ralf

Re: status on gdb-7.4 release...

[Jan Kratochvil]

On Thu, 22 Dec 2011 07:26:04 +0100, Joel Brobecker wrote:
>  2. 7.3->7.4 s390x regressions
>     http://www.sourceware.org/ml/gdb/2011-12/msg00039.html
> 
>     I think that these ones were marked as being a latent bug
>     caused by the kernel. Correct?
>     http://www.sourceware.org/ml/gdb-patches/2011-12/msg00743.html

Yes.  I have replied now to the mail above.


Thanks,
Jan

Re: status on gdb-7.4 release...

[Andreas Schwab]

Are these patches ok for the 7.4 branch?

http://sourceware.org/ml/gdb-patches/2011-12/msg00585.html
http://sourceware.org/ml/gdb-patches/2011-12/msg00451.html

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: status on gdb-7.4 release...

[Joel Brobecker]

> Are these patches ok for the 7.4 branch?
> 
> http://sourceware.org/ml/gdb-patches/2011-12/msg00585.html
> http://sourceware.org/ml/gdb-patches/2011-12/msg00451.html

Yes, they both look safe to me.

-- 
Joel

Re: status on gdb-7.4 release...

[Mark Kettenis]

> From: Andreas Schwab
> Date: Thu, 22 Dec 2011 11:03:17 +0100
> 
> Are these patches ok for the 7.4 branch?
> 
> http://sourceware.org/ml/gdb-patches/2011-12/msg00585.html
> http://sourceware.org/ml/gdb-patches/2011-12/msg00451.html

I'd say yes; first one only affects powerpc and the second is no-risk.

Re: status on gdb-7.4 release...

[Tom Tromey]

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

Joel> We had the few usual late discoveries and hicups, but the 7.4 release
Joel> seems well under way. As far as I know, there are a couple of
Joel> regressions that have been detected and still not fixed:

Joel>  1. Regression gdb.base/fixsection.exp && gdb.threads/thread_check.exp
Joel>     http://www.sourceware.org/ml/gdb-patches/2011-12/msg00560.html

I probably won't be able to fix this until early January.

Tom