From: Borislav Petkov <bp@alien8.de>
To: X86 ML <x86@kernel.org>
Cc: Andy Lutomirski <luto@amacapital.net>,
LKML <linux-kernel@vger.kernel.org>
Subject: [PATCH v2 15/15] x86/lib/memcpy_64.S: Convert memcpy to ALTERNATIVE_2 macro
Date: Tue, 24 Feb 2015 12:14:57 +0100 [thread overview]
Message-ID: <1424776497-3180-16-git-send-email-bp@alien8.de> (raw)
In-Reply-To: <1424776497-3180-1-git-send-email-bp@alien8.de>
From: Borislav Petkov <bp@suse.de>
Make REP_GOOD variant the default after alternatives have run.
Signed-off-by: Borislav Petkov <bp@suse.de>
---
arch/x86/lib/memcpy_64.S | 68 +++++++++++++++---------------------------------
1 file changed, 21 insertions(+), 47 deletions(-)
diff --git a/arch/x86/lib/memcpy_64.S b/arch/x86/lib/memcpy_64.S
index bbfdacc01760..b046664f5a1c 100644
--- a/arch/x86/lib/memcpy_64.S
+++ b/arch/x86/lib/memcpy_64.S
@@ -1,12 +1,20 @@
/* Copyright 2002 Andi Kleen */
#include <linux/linkage.h>
-
#include <asm/cpufeature.h>
#include <asm/dwarf2.h>
#include <asm/alternative-asm.h>
/*
+ * We build a jump to memcpy_orig by default which gets NOPped out on
+ * the majority of x86 CPUs which set REP_GOOD. In addition, CPUs which
+ * have the enhanced REP MOVSB/STOSB feature (ERMS), change those NOPs
+ * to a jmp to memcpy_erms which does the REP; MOVSB mem copy.
+ */
+
+.weak memcpy
+
+/*
* memcpy - Copy a memory block.
*
* Input:
@@ -17,15 +25,11 @@
* Output:
* rax original destination
*/
+ENTRY(__memcpy)
+ENTRY(memcpy)
+ ALTERNATIVE_2 "jmp memcpy_orig", "", X86_FEATURE_REP_GOOD, \
+ "jmp memcpy_erms", X86_FEATURE_ERMS
-/*
- * memcpy_c() - fast string ops (REP MOVSQ) based variant.
- *
- * This gets patched over the unrolled variant (below) via the
- * alternative instructions framework:
- */
- .section .altinstr_replacement, "ax", @progbits
-.Lmemcpy_c:
movq %rdi, %rax
movq %rdx, %rcx
shrq $3, %rcx
@@ -34,29 +38,21 @@
movl %edx, %ecx
rep movsb
ret
-.Lmemcpy_e:
- .previous
+ENDPROC(memcpy)
+ENDPROC(__memcpy)
/*
- * memcpy_c_e() - enhanced fast string memcpy. This is faster and simpler than
- * memcpy_c. Use memcpy_c_e when possible.
- *
- * This gets patched over the unrolled variant (below) via the
- * alternative instructions framework:
+ * memcpy_erms() - enhanced fast string memcpy. This is faster and
+ * simpler than memcpy. Use memcpy_erms when possible.
*/
- .section .altinstr_replacement, "ax", @progbits
-.Lmemcpy_c_e:
+ENTRY(memcpy_erms)
movq %rdi, %rax
movq %rdx, %rcx
rep movsb
ret
-.Lmemcpy_e_e:
- .previous
-
-.weak memcpy
+ENDPROC(memcpy_erms)
-ENTRY(__memcpy)
-ENTRY(memcpy)
+ENTRY(memcpy_orig)
CFI_STARTPROC
movq %rdi, %rax
@@ -183,26 +179,4 @@ ENTRY(memcpy)
.Lend:
retq
CFI_ENDPROC
-ENDPROC(memcpy)
-ENDPROC(__memcpy)
-
- /*
- * Some CPUs are adding enhanced REP MOVSB/STOSB feature
- * If the feature is supported, memcpy_c_e() is the first choice.
- * If enhanced rep movsb copy is not available, use fast string copy
- * memcpy_c() when possible. This is faster and code is simpler than
- * original memcpy().
- * Otherwise, original memcpy() is used.
- * In .altinstructions section, ERMS feature is placed after REG_GOOD
- * feature to implement the right patch order.
- *
- * Replace only beginning, memcpy is used to apply alternatives,
- * so it is silly to overwrite itself with nops - reboot is the
- * only outcome...
- */
- .section .altinstructions, "a"
- altinstruction_entry __memcpy,.Lmemcpy_c,X86_FEATURE_REP_GOOD,\
- .Lmemcpy_e-.Lmemcpy_c,.Lmemcpy_e-.Lmemcpy_c,0
- altinstruction_entry __memcpy,.Lmemcpy_c_e,X86_FEATURE_ERMS, \
- .Lmemcpy_e_e-.Lmemcpy_c_e,.Lmemcpy_e_e-.Lmemcpy_c_e,0
- .previous
+ENDPROC(memcpy_orig)
--
2.2.0.33.gc18b867
next prev parent reply other threads:[~2015-02-24 11:15 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-02-24 11:14 [PATCH v2 00/15] x86, alternatives: Instruction padding and more robust JMPs Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 01/15] x86/lib/copy_user_64.S: Remove FIX_ALIGNMENT define Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 02/15] x86/alternatives: Cleanup DPRINTK macro Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 03/15] x86/alternatives: Add instruction padding Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 04/15] x86/alternatives: Make JMPs more robust Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 05/15] x86/alternatives: Use optimized NOPs for padding Borislav Petkov
2015-03-04 6:43 ` Ingo Molnar
2015-03-04 8:42 ` Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 06/15] x86/lib/copy_page_64.S: Use generic ALTERNATIVE macro Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 07/15] x86/lib/copy_user_64.S: Convert to ALTERNATIVE_2 Borislav Petkov
2015-03-04 6:25 ` Ingo Molnar
2015-03-04 7:13 ` Ingo Molnar
2015-03-04 9:06 ` Borislav Petkov
2015-03-05 0:34 ` Ingo Molnar
2015-03-05 8:23 ` Borislav Petkov
2015-03-04 9:00 ` Borislav Petkov
2015-03-05 0:32 ` Ingo Molnar
2015-03-05 8:35 ` Borislav Petkov
2015-03-05 9:34 ` Ingo Molnar
2015-03-05 9:46 ` Ingo Molnar
2015-02-24 11:14 ` [PATCH v2 08/15] x86/smap: Use ALTERNATIVE macro Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 09/15] x86/entry_32: Convert X86_INVD_BUG to " Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 10/15] x86/lib/clear_page_64.S: Convert to ALTERNATIVE_2 macro Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 11/15] x86/asm: Use alternative_2() in rdtsc_barrier() Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 12/15] x86/asm: Cleanup prefetch primitives Borislav Petkov
2015-03-04 6:48 ` Ingo Molnar
2015-03-04 9:08 ` Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 13/15] x86/lib/memset_64.S: Convert to ALTERNATIVE_2 macro Borislav Petkov
2015-02-24 11:14 ` [PATCH v2 14/15] x86/lib/memmove_64.S: Convert memmove() to ALTERNATIVE macro Borislav Petkov
2015-03-04 7:19 ` Ingo Molnar
2015-02-24 11:14 ` Borislav Petkov [this message]
2015-03-04 7:26 ` [PATCH v2 15/15] x86/lib/memcpy_64.S: Convert memcpy to ALTERNATIVE_2 macro Ingo Molnar
2015-03-04 13:58 ` Borislav Petkov
2015-03-05 0:26 ` Ingo Molnar
2015-03-05 8:37 ` Borislav Petkov
2015-02-24 20:25 ` [PATCH v2 00/15] x86, alternatives: Instruction padding and more robust JMPs Andy Lutomirski
2015-02-26 18:13 ` Borislav Petkov
2015-02-26 18:16 ` [PATCH 1/3] perf/bench: Fix mem* routines usage after alternatives change Borislav Petkov
2015-02-26 18:16 ` [PATCH 2/3] perf/bench: Carve out mem routine benchmarking Borislav Petkov
2015-02-26 18:16 ` [PATCH 3/3] perf/bench: Add -r all so that you can run all mem* routines Borislav Petkov
2015-03-04 7:30 ` Ingo Molnar
2015-03-02 14:51 ` [PATCH v2 00/15] x86, alternatives: Instruction padding and more robust JMPs Hitoshi Mitake
2015-03-02 16:27 ` Borislav Petkov
2015-03-27 14:58 [PATCH v2 15/15] x86/lib/memcpy_64.S: Convert memcpy to ALTERNATIVE_2 macro David Laight
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=1424776497-3180-16-git-send-email-bp@alien8.de \
--to=bp@alien8.de \
--cc=linux-kernel@vger.kernel.org \
--cc=luto@amacapital.net \
--cc=x86@kernel.org \
/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
Powered by JetHome