From: "Måns Rullgård" <mans@mansr.com>
To: Richard Henderson <rth@twiddle.net>
Cc: libc-ports@sourceware.org, Joseph Myers <joseph@codesourcery.com>
Subject: Re: [PATCH 20/26] arm: Implement armv6t2 optimized strlen
Date: Wed, 27 Feb 2013 17:12:00 -0000 [thread overview]
Message-ID: <yw1xlia98yvl.fsf@unicorn.mansr.com> (raw)
In-Reply-To: <1361934986-17018-21-git-send-email-rth@twiddle.net> (Richard Henderson's message of "Tue, 26 Feb 2013 19:16:20 -0800")
Richard Henderson <rth@twiddle.net> writes:
> +ENTRY(strlen)
> + @ r0 = start of string
> + pld [r0]
> +
> + @ To cater to long strings, we want to search through a few
> + @ characters until we reach an aligned pointer. To cater to
> + @ small strings, we don't want to start doing word operations
> + @ immediately. The compromise is a maximum of 16 bytes less
> + @ whatever is required to end with an aligned pointer.
> + @ r3 = number of characters to search in alignment loop
> + and r3, r0, #7
> + s(mov) r1, r0 @ Save the input pointer
> + rsb r3, r3, #16
> +
> + @ Loop until we find ...
> +1: ldrb r2, [r0], #1
> + subs r3, r3, #1 @ ... the aligment point
> + it ne
> + cmpne r2, #0 @ ... or EOS
> + bne 1b
> +
> + @ Disambiguate the exit possibilites above
> + cmp r2, #0 @ Found EOS
> + ittt eq
> + subeq r0, r0, #1 @ Undo post-inc above
> + subeq r0, r0, r1 @ Subtract input to compute length
> + bxeq lr
> +
> + @ So now we're aligned.
> + ldrd r2, r3, [r0], #8
> + movw ip, #0xfefe
> + pld [r0, #64]
> + movt ip, #0xfefe
> + pld [r0, #128]
> + pld [r0, #192]
> +
> + @ Loop searching for EOS or C, 8 bytes at a time.
This comment seems to be for strchr().
> + @ Adding (unsigned saturating) 0xfe means result of 0xfe for any byte
> + @ that was originally zero and 0xff otherwise. Therefore we consider
> + @ the lsb of each byte the "found" bit, with 0 for a match.
> + .balign 16
> +2: uqadd8 r2, r2, ip @ Find EOS
> + uqadd8 r3, r3, ip
> + pld [r0, #256] @ Prefetch 4 lines ahead
> + s(and) r3, r3, r2 @ Combine the two words
> + mvns r3, r3 @ Test for any found bit true
> + it eq
> + ldrdeq r2, r3, [r0], #8
> + beq 2b
Subtracting the values (with UQSUB8) from 1 instead would result in a 0
result any non-zero input and a 1 for "found", i.e. the inverse of what
you have here. Testing for a match anywhere in the double-word then
becomes a single ORRS instruction. Unless I'm making some stupid mistake.
> + @ Found something. Disambiguate between first and second words.
> + @ Adjust r0 to point to the word containing the match.
> + @ Adjust r2 to the found bits for the word containing the match.
> + mvns r2, r2
> + itee ne
> + subne r0, r0, #8
> + moveq r2, r3
> + subeq r0, r0, #4
> +
> + @ Find the bit-offset of the match within the word.
> +#ifdef __ARMEL__
> + rbit r2, r2 @ For LE we need count-trailing-zeros
> +#endif
> + clz r2, r2
> + add r0, r0, r2, lsr #3 @ Adjust the pointer to the found byte
> + s(sub) r0, r0, r1 @ Subtract input to compute length
> + bx lr
> +
> +END(strlen)
This code could be made to work for any ARMv6 by (conditionally)
replacing the MOVW/MOVT with some equivalent and the RBIT by REV. REV
works since only the lsb in each byte can be set, so the result of CLZ
will simply be 7 more than we want, and the 3 low-order bits are shifted
out anyway.
--
Måns Rullgård
mans@mansr.com
next prev parent reply other threads:[~2013-02-27 17:12 UTC|newest]
Thread overview: 63+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-02-27 3:16 [PATCH 00/26] ARM improvements Richard Henderson
2013-02-27 3:16 ` [PATCH 01/26] Sync config.guess and config.sub with upstream Richard Henderson
2013-02-27 17:03 ` Joseph S. Myers
2013-02-27 3:16 ` [PATCH 05/26] arm: Introduce thumb helpers s and pc_ofs Richard Henderson
2013-02-28 0:20 ` Joseph S. Myers
2013-02-28 0:36 ` Richard Henderson
2013-02-28 1:45 ` Måns Rullgård
2013-02-27 3:16 ` [PATCH 09/26] arm: Mark assembly files that will not use thumb mode Richard Henderson
2013-02-28 0:58 ` Joseph S. Myers
2013-02-27 3:16 ` [PATCH 04/26] arm: Include libc-do-syscall in sysdep-rtld-routines Richard Henderson
2013-02-28 0:15 ` Joseph S. Myers
2013-02-27 3:16 ` [PATCH 03/26] arm: Handle armv6 in preconfigure Richard Henderson
2013-02-27 18:02 ` Joseph S. Myers
2013-02-27 18:04 ` Roland McGrath
2013-02-27 18:08 ` Richard Henderson
2013-02-27 18:34 ` Richard Henderson
2013-02-27 19:57 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 21/26] arm: Implement armv6t2 optimized strcpy Richard Henderson
2013-02-27 3:17 ` [PATCH 24/26] arm: Add optimized addmul_1 Richard Henderson
2013-02-28 13:58 ` Måns Rullgård
2013-02-28 18:19 ` Richard Henderson
2013-02-28 19:37 ` Måns Rullgård
2013-02-27 3:17 ` [PATCH 22/26] arm: Implement armv6t2 optimized strchr, strrchr, rawmemchr Richard Henderson
2013-02-28 1:31 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 26/26] arm: Add optimized add_n and sub_n Richard Henderson
2013-02-27 3:17 ` [PATCH 06/26] arm: Use pc_ofs Richard Henderson
2013-02-28 0:21 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 10/26] arm: Introduce and use LDST_PCREL Richard Henderson
2013-02-28 1:00 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 25/26] arm: Add optimized submul_1 Richard Henderson
2013-02-27 3:17 ` [PATCH 02/26] arm: Update preconfigure fragment for gcc 4.8 Richard Henderson
2013-02-27 17:54 ` Joseph S. Myers
2013-02-27 18:11 ` Richard Henderson
2013-02-27 3:17 ` [PATCH 12/26] arm: Enable thumb2 mode in assembly files Richard Henderson
2013-02-27 3:17 ` [PATCH 15/26] arm: Delete LOADREGS macro Richard Henderson
2013-02-28 1:24 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 16/26] arm: Commonize BX conditionals Richard Henderson
2013-02-28 21:51 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 08/26] arm: Add IT insns for thumb mode Richard Henderson
2013-02-28 0:41 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 11/26] arm: Introduce and use NEGOFF series of macros Richard Henderson
2013-02-27 3:17 ` [PATCH 18/26] arm: Use GET_TLS more often Richard Henderson
2013-02-28 21:59 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 13/26] arm: Store lr in r2 around GET_TLS Richard Henderson
2013-02-27 3:17 ` [PATCH 19/26] arm: Add optimized ffs for armv6t2 Richard Henderson
2013-02-27 15:51 ` Måns Rullgård
2013-02-27 16:34 ` Richard Henderson
2013-02-27 17:49 ` Roland McGrath
2013-02-27 3:17 ` [PATCH 07/26] arm: Introduce and use GET_TLS Richard Henderson
2013-02-28 0:34 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 14/26] arm: Use push/pop mnemonics Richard Henderson
2013-02-28 1:03 ` Joseph S. Myers
2013-02-27 3:17 ` [PATCH 23/26] arm: Rewrite armv6t2 memchr with uqadd8 Richard Henderson
2013-02-27 7:04 ` Richard Henderson
2013-02-27 3:17 ` [PATCH 20/26] arm: Implement armv6t2 optimized strlen Richard Henderson
2013-02-27 17:12 ` Måns Rullgård [this message]
2013-02-27 17:44 ` Richard Henderson
2013-02-27 3:17 ` [PATCH 17/26] arm: Unless arm4t, pop return address directly into pc Richard Henderson
2013-02-28 21:57 ` Joseph S. Myers
2013-02-27 15:41 ` [PATCH 00/26] ARM improvements Måns Rullgård
2013-02-27 16:59 ` Joseph S. Myers
2013-02-27 17:34 ` Richard Henderson
2013-02-28 22:05 ` Joseph S. Myers
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=yw1xlia98yvl.fsf@unicorn.mansr.com \
--to=mans@mansr.com \
--cc=joseph@codesourcery.com \
--cc=libc-ports@sourceware.org \
--cc=rth@twiddle.net \
/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 read-only IMAP folder(s) and NNTP newsgroup(s).