Re: [PATCH] Fix bootstrap failure on i686-pc-linux-gnu

[Caroline Tice <ctice at apple dot com>]

I don't know what the best thing to do would be.  I have updated the 
patch I
submitted last night to deal with Richard Henderson's comment (and
Geoff's patch) and I'm in the middle of running tests on it. The first
profiledbootstrap appears to be working, so I was about to submit it for
approval.  On the other hand it will take another day to finish running 
all
the tests...on the third hand :-) if we revert the patch from April 9, 
then I
have to re-create the patch all over and start the tests all over again.

I'll do whatever people think should be done at this point.  Comments
please?

-- Caroline Tice
ctice@apple.com


On Apr 12, 2005, at 12:18 PM, Kazu Hirata wrote:

> Hi Caroline,
>
> >          * dwarf2out.c (dwarf2out_switch_text_section): Check
> > current_function_decl
> >          and DECL_STRUCT_FUNCTION (current_function_decl) to make sure
> > they
> >          exist before trying to use them to access cold text section
> > name.
> >          (output_line_info): Likewise.
> >          (add_location_or_const_value_attribute): Likewise.
> >          (dwarf2out_var_location): Likewise.
> >          * varasm.c (initialize_cold_section_name): Likewise.
> >          (unlikely_text_section): Likewise.
> >          (in_unlikely_text_section): Likewise.
> >          (named_section): Test for cfun before trying to dereference 
> > it.
> >          (default_section_type_flags_1): Likewise.
>
> Could we revert your change on April 9 so that
>
> 1) you can take your time to adjust your patch to reflect Richard
>    Henderson's comment and test your patch, and
>
> 2) other people on i686-pc-linux-gnu can do usual development work?
>
> Mainline has been broken for almost three days.
>
> Kazu Hirata