Re: [PATCH][4/4][v2] RPO-style value-numbering for FRE/PRE

[Jeff Law <law@redhat.com>]

On 09/04/2018 04:12 PM, Gerald Pfeifer wrote:
> On Fri, 24 Aug 2018, Richard Biener wrote:
>> Comments are still welcome - I've re-bootstrapped and tested the series
>> on x86_64-unknown-linux-gnu for all languages and will talk about
>> this work at the Cauldron in more detail.
> 
> Is there any chance you can test this on i586 as well?
> 
> Since around that commit (August 27th) my i586 builds are failing
> with something like
> 
>   during GIMPLE pass: pre
>   cp-demangle.c: In function Β‘d_print_compΒ’:
>   cp-demangle.c:5711:1: internal compiler error: Segmentation fault
>   5711 | d_print_comp (struct d_print_info *dpi, int options,
>        | ^~~~~~~~~~~~
> 
> when doing a regular bootstrap on i586-unknown-freebsd10.4 (with
> clang 3.4.1 as system compiler).
> 
> Alas, when I add -save-temps, the internal compiler error goes away,
> and the backtrace using gdb installed on that system I share isn't 
> very useful, either.  (When I replace cp-demangle.c by cp-demangle.i
> in the invocation the error also disppears.)
> 
> On the other hand, this ICE has been consistent across a week of
> daily builds now.
An FYI, My i686 builds have been running OK.  But given what you've
described this could well be an uninitialized read, dangling pointer,
out of bounds write or some similar kind of bug.

Jeff