From: David Turner <dturner@twopensource.com>
To: "Torsten Bögershausen" <tboegi@web.de>
Cc: git@vger.kernel.org, gitster@pobox.com, mhagger@alum.mit.edu,
David Turner <dturner@twitter.com>
Subject: Re: [PATCH v6 2/2] refs.c: SSE4.2 optimizations for check_refname_component
Date: Wed, 04 Jun 2014 17:14:36 -0400 [thread overview]
Message-ID: <1401916476.18134.165.camel@stross> (raw)
In-Reply-To: <538ED2F1.9030003@web.de>
On Wed, 2014-06-04 at 10:04 +0200, Torsten Bögershausen wrote:
[snip discussion of compiler flags; I'll look into a cpuid approach]
> > --- a/git-compat-util.h
> > +++ b/git-compat-util.h
> > @@ -668,6 +668,28 @@ void git_qsort(void *base, size_t nmemb, size_t size,
> > #endif
> > #endif
> >
> > +#ifndef NO_SSE42
> > +#include <nmmintrin.h>
> > +/*
> > + * Clang ships with a version of nmmintrin.h that's incomplete; if
> > + * necessary, we define the constants that we're going to use.
> > + */
> > +#ifndef _SIDD_UBYTE_OPS
> > +#define _SIDD_UBYTE_OPS 0x00
> > +#define _SIDD_CMP_EQUAL_ANY 0x00
> > +#define _SIDD_CMP_RANGES 0x04
> > +#define _SIDD_CMP_EQUAL_ORDERED 0x0c
> > +#define _SIDD_NEGATIVE_POLARITY 0x10
> > +#endif
> Why do this defines end up in git-compat-util.h when they are needed by one file?
> (see even below)
Because Junio told me to:
"We would prefer not to add inclusion of any system header files in
random *.c files, as there often are system dependencies (order of
inclusion, definition of feature macros, etc.) we would rather want
to encapsulate in one place, that is git-compat-util.h."
> > --- a/refs.c
> > +++ b/refs.c
> > @@ -24,6 +24,25 @@ static unsigned char refname_disposition[256] = {
> > 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0, 0, 4, 4
> > };
> >
> > +static int check_refname_component_trailer(const char *cp, const char *refname, int flags)
> > +{
> > + if (cp == refname)
> > + return 0; /* Component has zero length. */
> > + if (refname[0] == '.') {
> > + if (!(flags & REFNAME_DOT_COMPONENT))
> > + return -1; /* Component starts with '.'. */
> > + /*
> > + * Even if leading dots are allowed, don't allow "."
> > + * as a component (".." is prevented by a rule above).
> > + */
> > + if (refname[1] == '\0')
> > + return -1; /* Component equals ".". */
> > + }
> > + if (cp - refname >= 5 && !memcmp(cp - 5, ".lock", 5))
> > + return -1; /* Refname ends with ".lock". */
> > + return cp - refname;
> > +}
> > +
> > /*
> > * Try to read one refname component from the front of refname.
> > * Return the length of the component found, or -1 if the component is
> > @@ -37,7 +56,7 @@ static unsigned char refname_disposition[256] = {
> > * - it ends with ".lock"
> > * - it contains a "\" (backslash)
> > */
> > -static int check_refname_component(const char *refname, int flags)
> > +static int check_refname_component_1(const char *refname, int flags)
> The name check_refname_component_1() doesn't tell too much,
> (check_refname_component_sse42() or check_refname_component_nonsse42() say more)
I'll go with "_bytewise", since that's how it works.
> can I suggest to move all SSE code out to a file under compat/,
> like compat/refs_sse42.c, or something similar ?
Since this is a relatively small section of code, I think that would be
overkill. Does anyone else have an opinion?
next prev parent reply other threads:[~2014-06-04 21:14 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-06-04 3:38 [PATCH v6 1/2] refs.c: optimize check_refname_component() David Turner
2014-06-04 3:38 ` [PATCH v6 2/2] refs.c: SSE4.2 optimizations for check_refname_component David Turner
2014-06-04 8:04 ` Torsten Bögershausen
2014-06-04 11:21 ` Duy Nguyen
2014-06-04 14:25 ` Torsten Bögershausen
2014-06-04 21:16 ` David Turner
2014-06-05 12:30 ` Torsten Bögershausen
2014-06-05 12:58 ` Ondřej Bílka
2014-06-05 19:26 ` David Turner
2014-06-05 21:42 ` Torsten Bögershausen
2014-06-05 22:02 ` David Turner
2014-06-04 21:14 ` David Turner [this message]
2014-06-04 21:46 ` Junio C Hamano
2014-06-05 19:27 ` David Turner
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1401916476.18134.165.camel@stross \
--to=dturner@twopensource.com \
--cc=dturner@twitter.com \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=mhagger@alum.mit.edu \
--cc=tboegi@web.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).