Re: v13: Performance regression related to FORTIFY_SOURCE - Mailing list pgsql-hackers

From Andres Freund
Subject Re: v13: Performance regression related to FORTIFY_SOURCE
Date
Msg-id 20200606024501.dnp3qqprnqu5dd24@alap3.anarazel.de
Whole thread Raw
In response to Re: v13: Performance regression related to FORTIFY_SOURCE  (Jeff Davis <pgsql@j-davis.com>)
List pgsql-hackers
Hi,

On 2020-06-05 18:39:28 -0700, Jeff Davis wrote:
> On Fri, 2020-06-05 at 14:49 -0700, Andres Freund wrote:
> > FWIW, with gcc 10 and glibc 2.30 I don't see such a switch. Taking a
> > profile shows me:
> 
> ...
> 
> >   4.65 │       → callq        memcpy@plt
> >        │       LogicalTapeWrite():
> > 
> > I.e. normal memcpy is getting called.
> > 
> > That's with -D_FORTIFY_SOURCE=2
> 
> That's good news, although people will be using ubuntu 18.04 for a
> while.
> 
> Just to confirm, would you mind trying the example programs in the GCC
> bug report?
> 
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95556

I get "call memcpy@PLT" for both files. With various debian versions of
gcc (7,8,9,10). But, very curiously, I do see the difference when
compiling with gcc-snapshot (which is a debian package wrapping a recent
snapshot from upstream gcc).

Greetings,

Andres Freund



pgsql-hackers by date:

Previous
From: Andres Freund
Date:
Subject: Re: Atomic operations within spinlocks
Next
From: Tom Lane
Date:
Subject: Re: Atomic operations within spinlocks