Re: [RFC PATCH v3 2/3] grep: make PCRE2 aware of custom allocator

From: René Scharfe <l.s.r@web.de>
Date: Aug 8, 2019, 14:29 UTC
Message-ID: <7f42007f-911b-c570-17f6-1c6af0429586@web.de>
In-Reply-To: <CAPUEspi2vu=7uLU19c-wcW+ji63gyeO7poAwS5_7ZhTiBWSxjg@mail.gmail.com>

Am 08.08.19 um 14:38 schrieb Carlo Arenas:

Show 24 quoted lines

> On Thu, Aug 8, 2019 at 12:07 AM René Scharfe <l.s.r@web.de> wrote:
>>
>> Am 08.08.19 um 04:35 schrieb Carlo Arenas:
>>> On Wed, Aug 7, 2019 at 6:03 AM René Scharfe <l.s.r@web.de> wrote:
>>>>
>>>> Am 07.08.19 um 11:49 schrieb Carlo Arenas:
>>>>> was hoping will perform better but it seems that testing can be done
>>>>> only in windows
>>>>
>>>> nedmalloc works on other platforms as well.
>>>
>>> I meant[1] it works reliably enough to be useful for performance testing.
>>
>> You mentioned being concerned about performance several times and I
>> wondered why each time.  I'd expect no measurable difference between
>> using a custom global context and the internal one of PCRE2 -- setting
>> two function pointers surely can't take very long, can it?  But
>> measuring is better than guessing, of course.
>
> setting the allocator is not a concern, but using it; it requires an
> extra indirect function call which is usually not very friendly to
> caches in our speculative execution CPU world.  our implementation
> also adds the wrapper call overhead, but in this case it is just the
> "cost of doing business" with PCRE2.

PCRE2 needs to allocate memory once per program run for the character table and for each pattern compilation. These are both rare events compared to matching patterns against lines, and I suspect that compilation in particular has much more other work to do, even more so with JIT enabled; I'd expect function call indirection to not make much of a difference.

pcre2_compile() always calls the allocation function through a function pointer in a context struct, by the way (see line 9695 in https://vcs.pcre.org/pcre2/code/trunk/src/pcre2_compile.c?view=markup or search for "malloc" in that file).

Show 11 quoted lines

>>> goes without saying that the fact that I am using a virtualbox with 2
>>> CPUs running Debian 10 on top of macOS (a macbook pro with 4 cores)
>>> and the test uses by default 8 threads, doesn't help,
>>
>> nedmalloc is supposed to run on macOS as well.
>
> the last version has some "fix miscompilations in macOS" fixes that
> might be relevant, and the version we have in tree says it works in
> the 32-bit version which latest macOS versions are working hard to
> deprecate (can't even build for it anymore), eitherway trying to run
> with a nedmalloc enabled git in macOS is not fun.

Importing the latest version of nedmalloc might make sense in general. The last commit in git://github.com/ned14/nedmalloc.git was done five years ago; is it finished? A diffstat with -b looks like this:

 malloc.c.h  | 1193 ++++++++++++++++++++++++++++-------------
 nedmalloc.c | 1720 +++++++++++++++++++++++++++++++++++++++++++++++++++++-------
 nedmalloc.h | 1580 ++++++++++++++++++++++++++++++++++++++++++++++++++++---
 3 files changed, 3840 insertions(+), 653 deletions(-)

Any nedmalloc fans interested in bringing the goodies hidden in there to Git (presumably while retaining our local fixes)?

Show 7 quoted lines

>> In particular I don't think that these results justify coupling the use
>> of nedmalloc to the choice of using a custom global context for PCRE2.
>
> neither did I either, the only reason I am holding on fully enabling
> NED with PCRE2 in my series is just because I wan't to make sure we
> have identified the bug correctly and we are fixing it (specially
> since I can't reproduce it, and therefore neither debug it)

That's a good reason against #ifdefs in general. Sometimes they are unavoidable, but they can make maintenance a lot harder.

> sorry for not making that clear enough, and as I said before, if we
> keep seeing segfaults even after v4 then we will have to do that or I
> might need to do a quick run to the nearest microsoft store hoping for
> a distracted rep, instead.

Asking to buy a license for Windows Vista might cause quite a bit of a distraction in there -- Microsoft's support for that version ended two years ago. :) It still seems to be popular enough to be supported by Git for Windows, however.

(You could buy Windows 10 and probably get a downgrade right, but finding legit install media for Vista might be challenging.)

But I'd say do the easy thing: Custom global context for all.

Show 21 quoted lines

>> I'd expect:
>> - Without USE_NED_ALLOCATOR: xmalloc() should be used for all
>>   allocations, including for PCRE2.  Some special exceptions use
>>   malloc(3) directly, but for most uses we want the consistent
>>   out-of-memory handling that xmalloc() brings.
>
> that is already in v4 and would expect to carry it forward.  this is
> also what I had in mind when I said we will need some fixes on top of
> Dscho version if we give up with these.
>
>> - With USE_NED_ALLOCATOR: malloc() and xmalloc() use nedmalloc
>>   behind the scenes and free() is similarly overridden, so all
>>   allocations are affected.
>> - If USE_NED_ALLOCATOR performs worse than the system allocator on
>>   some system then it's the problem of those that turn on that flag.
>>
>> Makes sense?
>
> completely, but note also that Dscho version would make the
> performance impacts of using a custom allocator (if any) affect
> everyone using PCRE2.

Sounds fine to me, if the performance numbers don't take too much of a hit. I'd be surprised if the needle moved at all (ignoring noise).

René