[Bug binutils/25180] Slowness in cplus_demangle()

[tim.ruehsen at gmx dot de]

https://sourceware.org/bugzilla/show_bug.cgi?id=25180

--- Comment #8 from Tim Rühsen <tim.ruehsen at gmx dot de> ---
Hi Nick,

(In reply to Nick Clifton from comment #7)
> > the attached patch fixes the demangler slowness (see commit message for a
> > description).
> 
> Did you run the libiberty testsuite to see if it introduces any new failures
> ?

I did, no (new) errors. What is the trick to run the test suite with valgrind
(except from manually running each test) ?

> Also it would be a good idea to run the g++ testsuite as well, just in
> case...

What is your suggestion to do that with least effort ?

> Given that you check for "d->d_counting > 1", would it make more sense to
> make that field a boolean ?  If not, then should the limit be configurable
> somehow ?

No and no. Booleans can't be > 1 ;-)
Just joking, in that case I also have to change d->d_printing to boolean, which
is IMO an unrelated optimization / cleanup.

I do not see any benefits in configuring the limit.

> Other than that though the patch looks fine to me, but of course I cannot
> approve it.  Are you going to submit it to gcc-patches ?

I will, thanks !

-- 
You are receiving this mail because:
You are on the CC list for the bug.