Hi,
This patch adds option -greadable-dwarf. In absence of an DW_AT_comment
attribute, it sets the DW_AT_name attribute of dies that otherwise do not get
that attribute, to make it easier to figure out what the die is describing.
The option exports the names of artificial variables:
...
DIE
On Wed, 15 Aug 2018, Tom de Vries wrote:
> Hi,
>
> This patch adds option -greadable-dwarf. In absence of an DW_AT_comment
> attribute,
What's a DW_AT_comment attribute? I don't see this mentioned in the
patch.
> it sets the DW_AT_name attribute of dies that otherwise do not get
> that attrib
How about adding this name to a -dA comment instead of the actual dwarf?
On Tue, Aug 21, 2018, 12:59 AM Richard Biener wrote:
> On Wed, 15 Aug 2018, Tom de Vries wrote:
>
> > Hi,
> >
> > This patch adds option -greadable-dwarf. In absence of an DW_AT_comment
> > attribute,
>
> What's a DW_AT_co
On Tue, 21 Aug 2018, Jason Merrill wrote:
> How about adding this name to a -dA comment instead of the actual dwarf?
That would also work but it requires either sticking that name somewhere
in die_struct or keep a reference to the decl we created a DIE for
in the same struct. Though we already h
On 08/20/2018 02:59 PM, Richard Biener wrote:
> On Wed, 15 Aug 2018, Tom de Vries wrote:
>
>> Hi,
>>
>> This patch adds option -greadable-dwarf. In absence of an DW_AT_comment
>> attribute,
>
> What's a DW_AT_comment attribute? I don't see this mentioned in the
> patch.
>
It's a hypothetical
Hi,
That solution is limited: it does not show the extra information in the
dump from an executable (which is the only way to see how things look
like post-linking).
Is your concern leaking compiler internals into the dwarf info?
Thanks,
- Tom
On 08/21/2018 06:53 AM, Jason Merrill wrote:
> How
