Re: [PATCH 1/3] gdb/ctf: don't create a buildsym_compunit when building partial symbols

[Wei-min Pan via Gdb-patches <gdb-patches@sourceware.org>]

Hi Simon,

On 3/31/2022 5:14 PM, Simon Marchi wrote:
>> A CTF archive contains conflicting type info and is treated as a CU.
>> Since one can't switch scope among CUs in CTF, not reading archives
>> might not be a problem. Please go ahead and make the change. It can
>> be revisited when we come up a test case that shows otherwise.
>>
>> Thanks.
> Hi Wei-Min,
>
> I'm not sure we are on the same page.  I don't know much about CTF, so
> let's make sure I understand things right.  I am not sure how CTF debug
> info is organized, and in particular what an "archive" is.  If you can
> explain this or give a reference, that would be useful.

Here is the spec for CTF V3:

http://www.esperi.org.uk/~oranix/ctf/ctf-spec.pdf

which hopefully is helpful.

>
> I'll attach the output of "readelf --ctf=.ctf" for the file I am working
> with.
>
> My understanding from the readelf output is that there is some debug
> info (function, types, etc) described at the top-level.  It doesn't seem
> associated to a source file, and it seems to contain things that are
> global.
>
> Then there is one "archive" for each compiled .c file, such as:
>
>    CTF archive member: /home/simark/src/babeltrace/src/lib/current-thread.c:
>
> Each archive contains very little data, my guess is that it only
> contains what is not already described at the top-level?

Yes, that's my understanding too.

>
> Right now, in GDB, we create one CU (partial_symtab, and then
> compunit_symtab after expansion) for the top-level, and the one CU for
> each archive.  My patch does not change that.  It only removes a
> "new buildsym_compunit" call that just seems useless / to do nothing.
>
> In other words, no user-visible changes are expected from this patch.

OK. Thanks.

>
> Simon