patches for mac support

patches for mac support

[Fawzi Mohamed]

Hi all,

I have worked on bug http://sourceware.org/bugzilla/show_bug.cgi?id=11488 .
Now I think that I have reached to bottom of it, and I have a set of clean patches that could use a review.

Apple gdb simply ignores the eh_frame sections for the libraries without embedded dwarf info, not using eh information (which was the first patch that I submitted) but I find that one should be able to use them, so I looked further and found the "correct" fix.
As the path to it was a bit convoluted I have done a few improvements to pieces of code that did fail as consequence of the original bug.

The "main" fix is
   http://sourceware.org/bugzilla/attachment.cgi?id=5816&action=diff
which fixes the mmaped read of sections.

I feel that
	http://sourceware.org/bugzilla/attachment.cgi?id=5817
which adds a check on the cie pointer is important and improves gdb robustness and should also go in.

	http://sourceware.org/bugzilla/attachment.cgi?id=5814
adds the forgotten ignore of routine_64 load command, and should also go in

	http://sourceware.org/bugzilla/attachment.cgi?id=5809
is something that I have seen apple does and seems reasonable, but I have not needed it, so I am not sure if it should go in

I have also noted that compiling gdb I have the following conflict in the grammar, not sure if you are aware of it

/bin/sh ../../gdb72/gdb/../ylwrap ../../gdb72/gdb/m2-exp.y y.tab.c m2-exp.c.tmp -- bison -y
conflicts: 34 shift/reduce
/Users/fawzi/dev/gdb/gdb72Build/gdb/../../gdb72/gdb/m2-exp.y:355.25-44: warning: rule never reduced because of conflicts: @2: /* empty */

ciao
Fawzi

Re: patches for mac support

[Tristan Gingold]

On Jun 23, 2011, at 10:07 PM, Fawzi Mohamed wrote:

Mohamed,

First, thank you for this work.

> I have worked on bug http://sourceware.org/bugzilla/show_bug.cgi?id=11488 .
> Now I think that I have reached to bottom of it, and I have a set of clean patches that could use a review.
> 
> Apple gdb simply ignores the eh_frame sections for the libraries without embedded dwarf info, not using eh information (which was the first patch that I submitted) but I find that one should be able to use them, so I looked further and found the "correct" fix.
> As the path to it was a bit convoluted I have done a few improvements to pieces of code that did fail as consequence of the original bug.
> 
> The "main" fix is
>   http://sourceware.org/bugzilla/attachment.cgi?id=5816&action=diff
> which fixes the mmaped read of sections.

We (AdaCore) has something very similar in our internal tree.  I very recently worked on submitting a patch to fix this issue.
The submitted patch was larger as I tried to clean-up this API.

Note that you need to properly submit your patches: one patch by mail is better (IMHO), patch should be inlined if not large, and
you need ChangeLog entries.  I suppose the instructions are on the web, also I don't know where.

> I feel that
> 	http://sourceware.org/bugzilla/attachment.cgi?id=5817
> which adds a check on the cie pointer is important and improves gdb robustness and should also go in.
> 
> 	http://sourceware.org/bugzilla/attachment.cgi?id=5814
> adds the forgotten ignore of routine_64 load command, and should also go in

Note that this should be submitted on the binutils mailing list.

> 	http://sourceware.org/bugzilla/attachment.cgi?id=5809
> is something that I have seen apple does and seems reasonable, but I have not needed it, so I am not sure if it should go in
> 
> I have also noted that compiling gdb I have the following conflict in the grammar, not sure if you are aware of it
> 
> /bin/sh ../../gdb72/gdb/../ylwrap ../../gdb72/gdb/m2-exp.y y.tab.c m2-exp.c.tmp -- bison -y
> conflicts: 34 shift/reduce
> /Users/fawzi/dev/gdb/gdb72Build/gdb/../../gdb72/gdb/m2-exp.y:355.25-44: warning: rule never reduced because of conflicts: @2: /* empty */

Tristan.

Re: patches for mac support

[Pedro Alves]

On Friday 24 June 2011 10:38:08, Tristan Gingold wrote:

> Note that you need to properly submit your patches: one patch by mail is better (IMHO), patch should be inlined if not large, and
> you need ChangeLog entries.  I suppose the instructions are on the web, also I don't know where.

<http://www.gnu.org/software/gdb/>, follow [contributing] at the top.

-- 
Pedro Alves

RE: patches for mac support

[fawzi.mohamed]

On Friday, June 24, 2011 at 11:38 AM Tristan Gingold [gingold@adacore.com] wrote:

On Jun 23, 2011, at 10:07 PM, Fawzi Mohamed wrote:

> Mohamed,
> 
> First, thank you for this work.

you are welcome

> > I have worked on bug http://sourceware.org/bugzilla/show_bug.cgi?id=11488 .
> > Now I think that I have reached to bottom of it, and I have a set of clean patches that could use a review.
> >
> > Apple gdb simply ignores the eh_frame sections for the libraries without embedded dwarf info, not using eh information (which was the first patch that I submitted) but I find that one should be able to use them, so I looked further and found the "correct" fix.
> > As the path to it was a bit convoluted I have done a few improvements to pieces of code that did fail as consequence of the original bug.
> >
> > The "main" fix is
> >   http://sourceware.org/bugzilla/attachment.cgi?id=5816&action=diff
> > which fixes the mmaped read of sections.
> 
> We (AdaCore) has something very similar in our internal tree.  I very recently worked on submitting a patch to fix this issue.
> The submitted patch was larger as I tried to clean-up this API.

great, I don't care the way it goes, as long as it gets fixed...,
so I suppose I can skip submitting it, unless yours doesn't get accepted.
A pity hat you didn't tell this in the bug, I would have spared some debugging,
I am a lazy guy ;)...

> Note that you need to properly submit your patches: one patch by mail is better (IMHO), patch should be inlined if not large, and
> you need ChangeLog entries.  I suppose the instructions are on the web, also I don't know where.

Pedro Alves answered this, thanks, I am new to submitting to gdb

> > I feel that
> >       http://sourceware.org/bugzilla/attachment.cgi?id=5817
> > which adds a check on the cie pointer is important and improves gdb robustness and should also go in.
> >
> >       http://sourceware.org/bugzilla/attachment.cgi?id=5814
> > adds the forgotten ignore of routine_64 load command, and should also go in
> 
> Note that this should be submitted on the binutils mailing list.

ok I will do...

> >       http://sourceware.org/bugzilla/attachment.cgi?id=5809
> > is something that I have seen apple does and seems reasonable, but I have not needed it, so I am not sure if it should go in
> >
> > I have also noted that compiling gdb I have the following conflict in the grammar, not sure if you are aware of it
> >
> > /bin/sh ../../gdb72/gdb/../ylwrap ../../gdb72/gdb/m2-exp.y y.tab.c m2-exp.c.tmp -- bison -y
> > conflicts: 34 shift/reduce
> > /Users/fawzi/dev/gdb/gdb72Build/gdb/../../gdb72/gdb/m2-exp.y:355.25-44: warning: rule never reduced because of conflicts: @2: /* empty */
> 
> Tristan.

ciao
Fawzi

Re: patches for mac support

[Tristan Gingold]

On Jun 26, 2011, at 2:22 AM, <fawzi.mohamed@nokia.com> <fawzi.mohamed@nokia.com> wrote:

> On Friday, June 24, 2011 at 11:38 AM Tristan Gingold [gingold@adacore.com] wrote:
> 
> On Jun 23, 2011, at 10:07 PM, Fawzi Mohamed wrote:
> 
>> Mohamed,
>> 
>> First, thank you for this work.
> 
> you are welcome
> 
>>> I have worked on bug http://sourceware.org/bugzilla/show_bug.cgi?id=11488 .
>>> Now I think that I have reached to bottom of it, and I have a set of clean patches that could use a review.
>>> 
>>> Apple gdb simply ignores the eh_frame sections for the libraries without embedded dwarf info, not using eh information (which was the first patch that I submitted) but I find that one should be able to use them, so I looked further and found the "correct" fix.
>>> As the path to it was a bit convoluted I have done a few improvements to pieces of code that did fail as consequence of the original bug.
>>> 
>>> The "main" fix is
>>>  http://sourceware.org/bugzilla/attachment.cgi?id=5816&action=diff
>>> which fixes the mmaped read of sections.
>> 
>> We (AdaCore) has something very similar in our internal tree.  I very recently worked on submitting a patch to fix this issue.
>> The submitted patch was larger as I tried to clean-up this API.
> 
> great, I don't care the way it goes, as long as it gets fixed...,
> so I suppose I can skip submitting it, unless yours doesn't get accepted.
> A pity hat you didn't tell this in the bug, I would have spared some debugging,

I am a lazy guy too :-)

> I am a lazy guy ;)...

Anyway, I have just committed our patch.

BTW, do you (or Nokia) have the FSF copyright assignment ?

Tristan.

Re: patches for mac support

[Fawzi Mohamed]

On Jun 27, 2011, at 10:59 AM, ext Tristan Gingold wrote:

> 
> On Jun 26, 2011, at 2:22 AM, <fawzi.mohamed@nokia.com> <fawzi.mohamed@nokia.com> wrote:
> 
>> On Friday, June 24, 2011 at 11:38 AM Tristan Gingold [gingold@adacore.com] wrote:
>> 
>> On Jun 23, 2011, at 10:07 PM, Fawzi Mohamed wrote:
>> 
>>> Mohamed,
>>> 
>>> First, thank you for this work.
>> 
>> you are welcome
>> 
>>>> I have worked on bug http://sourceware.org/bugzilla/show_bug.cgi?id=11488 .
>>>> Now I think that I have reached to bottom of it, and I have a set of clean patches that could use a review.
>>>> 
>>>> Apple gdb simply ignores the eh_frame sections for the libraries without embedded dwarf info, not using eh information (which was the first patch that I submitted) but I find that one should be able to use them, so I looked further and found the "correct" fix.
>>>> As the path to it was a bit convoluted I have done a few improvements to pieces of code that did fail as consequence of the original bug.
>>>> 
>>>> The "main" fix is
>>>> http://sourceware.org/bugzilla/attachment.cgi?id=5816&action=diff
>>>> which fixes the mmaped read of sections.
>>> 
>>> We (AdaCore) has something very similar in our internal tree.  I very recently worked on submitting a patch to fix this issue.
>>> The submitted patch was larger as I tried to clean-up this API.
>> 
>> great, I don't care the way it goes, as long as it gets fixed...,
>> so I suppose I can skip submitting it, unless yours doesn't get accepted.
>> A pity hat you didn't tell this in the bug, I would have spared some debugging,
> 
> I am a lazy guy too :-)
> 
>> I am a lazy guy ;)...
> 
> Anyway, I have just committed our patch.
> 
> BTW, do you (or Nokia) have the FSF copyright assignment ?

yes Nokia has a global copyright assignment to FSF

ciao
Fawzi

Re: patches for mac support

[André Pönitz]

On Monday 27 June 2011 10:59:12 ext Tristan Gingold wrote:
> [...]
> > On Jun 23, 2011, at 10:07 PM, Fawzi Mohamed wrote:
> >  [...]
> >>> I have worked on bug http://sourceware.org/bugzilla/show_bug.cgi?id=11488 .
> >>> Now I think that I have reached to bottom of it, and I have a set of clean patches that could use a review.
> >>> 
> >>> Apple gdb simply ignores the eh_frame sections for the libraries without embedded dwarf info, not using eh information (which was the first patch that I submitted) but I find that one should be able to use them, so I looked further and found the "correct" fix.
> >>> As the path to it was a bit convoluted I have done a few improvements to pieces of code that did fail as consequence of the original bug.
> >>> 
> >>> The "main" fix is
> >>>  http://sourceware.org/bugzilla/attachment.cgi?id=5816&action=diff
> >>> which fixes the mmaped read of sections.
> >> 
> >> We (AdaCore) has something very similar in our internal tree.  I very recently worked on submitting a patch to fix this issue.
> >> The submitted patch was larger as I tried to clean-up this API.
> > 
> > great, I don't care the way it goes, as long as it gets fixed...,
> > so I suppose I can skip submitting it, unless yours doesn't get accepted.
> > A pity hat you didn't tell this in the bug, I would have spared some debugging,
> 
> I am a lazy guy too :-)
> 
> > I am a lazy guy ;)...
> 
> Anyway, I have just committed our patch.

Thanks a lot for the effort.

> BTW, do you (or Nokia) have the FSF copyright assignment ?

There should be a company-wide one for gdb. Extremely agile process btw.

But is that relevant at all if you apply _your_ patch?

Regards,
Andre'

Re: patches for mac support

[Fawzi Mohamed]

On Jun 24, 2011, at 11:38 AM, ext Tristan Gingold wrote:

> 
> On Jun 23, 2011, at 10:07 PM, Fawzi Mohamed wrote:
> 
> [...]
> Note that you need to properly submit your patches: one patch by mail is better (IMHO), patch should be inlined if not large, and
> you need ChangeLog entries.  I suppose the instructions are on the web, also I don't know where.
> 
>> I feel that
>> 	http://sourceware.org/bugzilla/attachment.cgi?id=5817
>> which adds a check on the cie pointer is important and improves gdb robustness and should also go in.

I have reformatted it, but I am waiting for a compiling trunk on mac before trying to submit it (the removal of ENUM_BITFIELD was not complete).

>> 
>> 	http://sourceware.org/bugzilla/attachment.cgi?id=5814
>> adds the forgotten ignore of routine_64 load command, and should also go in
> 
> Note that this should be submitted on the binutils mailing list.

done

Fawzi