Commit 346ce1d7 authored by Linus Torvalds's avatar Linus Torvalds

Merge branch 'parisc-4.11-3' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux

Pull parisc fixes from Helge Deller:
 "Al Viro reported that - in case of read faults - our copy_from_user()
  implementation may claim to have copied more bytes than it actually
  did. In order to fix this bug and because of the way how gcc optimizes
  register usage for inline assembly in C code, we had to replace our
  pa_memcpy() function with a pure assembler implementation.

  While fixing the memcpy bug we noticed some other issues with our
  get_user() and put_user() functions, e.g. nested faults may return
  wrong data. This is now fixed by a common fixup handler for
  get_user/put_user in the exception handler which additionally makes
  generated code smaller and faster.

  The third patch is a trivial one-line fix for a patch which went in
  during 4.11-rc and which avoids stalled CPU warnings after power
  shutdown (for parisc machines which can't plug power off themselves).

  Due to the rewrite of pa_memcpy() into assembly this patch got bigger
  than what I wanted to have sent at this stage.

  Those patches have been running in production during the last few days
  on our debian build servers without any further issues"

* 'parisc-4.11-3' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Avoid stalled CPU warnings after system shutdown
  parisc: Clean up fixup routines for get_user()/put_user()
  parisc: Fix access fault handling in pa_memcpy()
parents 7d34ddbe 476e75a4
......@@ -64,6 +64,15 @@ struct exception_table_entry {
".word (" #fault_addr " - .), (" #except_addr " - .)\n\t" \
".previous\n"
/*
* ASM_EXCEPTIONTABLE_ENTRY_EFAULT() creates a special exception table entry
* (with lowest bit set) for which the fault handler in fixup_exception() will
* load -EFAULT into %r8 for a read or write fault, and zeroes the target
* register in case of a read fault in get_user().
*/
#define ASM_EXCEPTIONTABLE_ENTRY_EFAULT( fault_addr, except_addr )\
ASM_EXCEPTIONTABLE_ENTRY( fault_addr, except_addr + 1)
/*
* The page fault handler stores, in a per-cpu area, the following information
* if a fixup routine is available.
......@@ -91,7 +100,7 @@ struct exception_data {
#define __get_user(x, ptr) \
({ \
register long __gu_err __asm__ ("r8") = 0; \
register long __gu_val __asm__ ("r9") = 0; \
register long __gu_val; \
\
load_sr2(); \
switch (sizeof(*(ptr))) { \
......@@ -107,22 +116,23 @@ struct exception_data {
})
#define __get_user_asm(ldx, ptr) \
__asm__("\n1:\t" ldx "\t0(%%sr2,%2),%0\n\t" \
ASM_EXCEPTIONTABLE_ENTRY(1b, fixup_get_user_skip_1)\
__asm__("1: " ldx " 0(%%sr2,%2),%0\n" \
"9:\n" \
ASM_EXCEPTIONTABLE_ENTRY_EFAULT(1b, 9b) \
: "=r"(__gu_val), "=r"(__gu_err) \
: "r"(ptr), "1"(__gu_err) \
: "r1");
: "r"(ptr), "1"(__gu_err));
#if !defined(CONFIG_64BIT)
#define __get_user_asm64(ptr) \
__asm__("\n1:\tldw 0(%%sr2,%2),%0" \
"\n2:\tldw 4(%%sr2,%2),%R0\n\t" \
ASM_EXCEPTIONTABLE_ENTRY(1b, fixup_get_user_skip_2)\
ASM_EXCEPTIONTABLE_ENTRY(2b, fixup_get_user_skip_1)\
__asm__(" copy %%r0,%R0\n" \
"1: ldw 0(%%sr2,%2),%0\n" \
"2: ldw 4(%%sr2,%2),%R0\n" \
"9:\n" \
ASM_EXCEPTIONTABLE_ENTRY_EFAULT(1b, 9b) \
ASM_EXCEPTIONTABLE_ENTRY_EFAULT(2b, 9b) \
: "=r"(__gu_val), "=r"(__gu_err) \
: "r"(ptr), "1"(__gu_err) \
: "r1");
: "r"(ptr), "1"(__gu_err));
#endif /* !defined(CONFIG_64BIT) */
......@@ -148,32 +158,31 @@ struct exception_data {
* The "__put_user/kernel_asm()" macros tell gcc they read from memory
* instead of writing. This is because they do not write to any memory
* gcc knows about, so there are no aliasing issues. These macros must
* also be aware that "fixup_put_user_skip_[12]" are executed in the
* context of the fault, and any registers used there must be listed
* as clobbers. In this case only "r1" is used by the current routines.
* r8/r9 are already listed as err/val.
* also be aware that fixups are executed in the context of the fault,
* and any registers used there must be listed as clobbers.
* r8 is already listed as err.
*/
#define __put_user_asm(stx, x, ptr) \
__asm__ __volatile__ ( \
"\n1:\t" stx "\t%2,0(%%sr2,%1)\n\t" \
ASM_EXCEPTIONTABLE_ENTRY(1b, fixup_put_user_skip_1)\
"1: " stx " %2,0(%%sr2,%1)\n" \
"9:\n" \
ASM_EXCEPTIONTABLE_ENTRY_EFAULT(1b, 9b) \
: "=r"(__pu_err) \
: "r"(ptr), "r"(x), "0"(__pu_err) \
: "r1")
: "r"(ptr), "r"(x), "0"(__pu_err))
#if !defined(CONFIG_64BIT)
#define __put_user_asm64(__val, ptr) do { \
__asm__ __volatile__ ( \
"\n1:\tstw %2,0(%%sr2,%1)" \
"\n2:\tstw %R2,4(%%sr2,%1)\n\t" \
ASM_EXCEPTIONTABLE_ENTRY(1b, fixup_put_user_skip_2)\
ASM_EXCEPTIONTABLE_ENTRY(2b, fixup_put_user_skip_1)\
"1: stw %2,0(%%sr2,%1)\n" \
"2: stw %R2,4(%%sr2,%1)\n" \
"9:\n" \
ASM_EXCEPTIONTABLE_ENTRY_EFAULT(1b, 9b) \
ASM_EXCEPTIONTABLE_ENTRY_EFAULT(2b, 9b) \
: "=r"(__pu_err) \
: "r"(ptr), "r"(__val), "0"(__pu_err) \
: "r1"); \
: "r"(ptr), "r"(__val), "0"(__pu_err)); \
} while (0)
#endif /* !defined(CONFIG_64BIT) */
......
......@@ -47,16 +47,6 @@ EXPORT_SYMBOL(__cmpxchg_u64);
EXPORT_SYMBOL(lclear_user);
EXPORT_SYMBOL(lstrnlen_user);
/* Global fixups - defined as int to avoid creation of function pointers */
extern int fixup_get_user_skip_1;
extern int fixup_get_user_skip_2;
extern int fixup_put_user_skip_1;
extern int fixup_put_user_skip_2;
EXPORT_SYMBOL(fixup_get_user_skip_1);
EXPORT_SYMBOL(fixup_get_user_skip_2);
EXPORT_SYMBOL(fixup_put_user_skip_1);
EXPORT_SYMBOL(fixup_put_user_skip_2);
#ifndef CONFIG_64BIT
/* Needed so insmod can set dp value */
extern int $global$;
......
......@@ -143,6 +143,8 @@ void machine_power_off(void)
printk(KERN_EMERG "System shut down completed.\n"
"Please power this system off now.");
/* prevent soft lockup/stalled CPU messages for endless loop. */
rcu_sysrq_start();
for (;;);
}
......
......@@ -2,7 +2,7 @@
# Makefile for parisc-specific library files
#
lib-y := lusercopy.o bitops.o checksum.o io.o memset.o fixup.o memcpy.o \
lib-y := lusercopy.o bitops.o checksum.o io.o memset.o memcpy.o \
ucmpdi2.o delay.o
obj-y := iomap.o
/*
* Linux/PA-RISC Project (http://www.parisc-linux.org/)
*
* Copyright (C) 2004 Randolph Chung <tausq@debian.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2, or (at your option)
* any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*
* Fixup routines for kernel exception handling.
*/
#include <asm/asm-offsets.h>
#include <asm/assembly.h>
#include <asm/errno.h>
#include <linux/linkage.h>
#ifdef CONFIG_SMP
.macro get_fault_ip t1 t2
loadgp
addil LT%__per_cpu_offset,%r27
LDREG RT%__per_cpu_offset(%r1),\t1
/* t2 = smp_processor_id() */
mfctl 30,\t2
ldw TI_CPU(\t2),\t2
#ifdef CONFIG_64BIT
extrd,u \t2,63,32,\t2
#endif
/* t2 = &__per_cpu_offset[smp_processor_id()]; */
LDREGX \t2(\t1),\t2
addil LT%exception_data,%r27
LDREG RT%exception_data(%r1),\t1
/* t1 = this_cpu_ptr(&exception_data) */
add,l \t1,\t2,\t1
/* %r27 = t1->fault_gp - restore gp */
LDREG EXCDATA_GP(\t1), %r27
/* t1 = t1->fault_ip */
LDREG EXCDATA_IP(\t1), \t1
.endm
#else
.macro get_fault_ip t1 t2
loadgp
/* t1 = this_cpu_ptr(&exception_data) */
addil LT%exception_data,%r27
LDREG RT%exception_data(%r1),\t2
/* %r27 = t2->fault_gp - restore gp */
LDREG EXCDATA_GP(\t2), %r27
/* t1 = t2->fault_ip */
LDREG EXCDATA_IP(\t2), \t1
.endm
#endif
.level LEVEL
.text
.section .fixup, "ax"
/* get_user() fixups, store -EFAULT in r8, and 0 in r9 */
ENTRY_CFI(fixup_get_user_skip_1)
get_fault_ip %r1,%r8
ldo 4(%r1), %r1
ldi -EFAULT, %r8
bv %r0(%r1)
copy %r0, %r9
ENDPROC_CFI(fixup_get_user_skip_1)
ENTRY_CFI(fixup_get_user_skip_2)
get_fault_ip %r1,%r8
ldo 8(%r1), %r1
ldi -EFAULT, %r8
bv %r0(%r1)
copy %r0, %r9
ENDPROC_CFI(fixup_get_user_skip_2)
/* put_user() fixups, store -EFAULT in r8 */
ENTRY_CFI(fixup_put_user_skip_1)
get_fault_ip %r1,%r8
ldo 4(%r1), %r1
bv %r0(%r1)
ldi -EFAULT, %r8
ENDPROC_CFI(fixup_put_user_skip_1)
ENTRY_CFI(fixup_put_user_skip_2)
get_fault_ip %r1,%r8
ldo 8(%r1), %r1
bv %r0(%r1)
ldi -EFAULT, %r8
ENDPROC_CFI(fixup_put_user_skip_2)
......@@ -5,6 +5,8 @@
* Copyright (C) 2000 Richard Hirst <rhirst with parisc-linux.org>
* Copyright (C) 2001 Matthieu Delahaye <delahaym at esiee.fr>
* Copyright (C) 2003 Randolph Chung <tausq with parisc-linux.org>
* Copyright (C) 2017 Helge Deller <deller@gmx.de>
* Copyright (C) 2017 John David Anglin <dave.anglin@bell.net>
*
*
* This program is free software; you can redistribute it and/or modify
......@@ -132,4 +134,320 @@ ENDPROC_CFI(lstrnlen_user)
.procend
/*
* unsigned long pa_memcpy(void *dstp, const void *srcp, unsigned long len)
*
* Inputs:
* - sr1 already contains space of source region
* - sr2 already contains space of destination region
*
* Returns:
* - number of bytes that could not be copied.
* On success, this will be zero.
*
* This code is based on a C-implementation of a copy routine written by
* Randolph Chung, which in turn was derived from the glibc.
*
* Several strategies are tried to try to get the best performance for various
* conditions. In the optimal case, we copy by loops that copy 32- or 16-bytes
* at a time using general registers. Unaligned copies are handled either by
* aligning the destination and then using shift-and-write method, or in a few
* cases by falling back to a byte-at-a-time copy.
*
* Testing with various alignments and buffer sizes shows that this code is
* often >10x faster than a simple byte-at-a-time copy, even for strangely
* aligned operands. It is interesting to note that the glibc version of memcpy
* (written in C) is actually quite fast already. This routine is able to beat
* it by 30-40% for aligned copies because of the loop unrolling, but in some
* cases the glibc version is still slightly faster. This lends more
* credibility that gcc can generate very good code as long as we are careful.
*
* Possible optimizations:
* - add cache prefetching
* - try not to use the post-increment address modifiers; they may create
* additional interlocks. Assumption is that those were only efficient on old
* machines (pre PA8000 processors)
*/
dst = arg0
src = arg1
len = arg2
end = arg3
t1 = r19
t2 = r20
t3 = r21
t4 = r22
srcspc = sr1
dstspc = sr2
t0 = r1
a1 = t1
a2 = t2
a3 = t3
a0 = t4
save_src = ret0
save_dst = ret1
save_len = r31
ENTRY_CFI(pa_memcpy)
.proc
.callinfo NO_CALLS
.entry
/* Last destination address */
add dst,len,end
/* short copy with less than 16 bytes? */
cmpib,>>=,n 15,len,.Lbyte_loop
/* same alignment? */
xor src,dst,t0
extru t0,31,2,t1
cmpib,<>,n 0,t1,.Lunaligned_copy
#ifdef CONFIG_64BIT
/* only do 64-bit copies if we can get aligned. */
extru t0,31,3,t1
cmpib,<>,n 0,t1,.Lalign_loop32
/* loop until we are 64-bit aligned */
.Lalign_loop64:
extru dst,31,3,t1
cmpib,=,n 0,t1,.Lcopy_loop_16
20: ldb,ma 1(srcspc,src),t1
21: stb,ma t1,1(dstspc,dst)
b .Lalign_loop64
ldo -1(len),len
ASM_EXCEPTIONTABLE_ENTRY(20b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(21b,.Lcopy_done)
ldi 31,t0
.Lcopy_loop_16:
cmpb,COND(>>=),n t0,len,.Lword_loop
10: ldd 0(srcspc,src),t1
11: ldd 8(srcspc,src),t2
ldo 16(src),src
12: std,ma t1,8(dstspc,dst)
13: std,ma t2,8(dstspc,dst)
14: ldd 0(srcspc,src),t1
15: ldd 8(srcspc,src),t2
ldo 16(src),src
16: std,ma t1,8(dstspc,dst)
17: std,ma t2,8(dstspc,dst)
ASM_EXCEPTIONTABLE_ENTRY(10b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(11b,.Lcopy16_fault)
ASM_EXCEPTIONTABLE_ENTRY(12b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(13b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(14b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(15b,.Lcopy16_fault)
ASM_EXCEPTIONTABLE_ENTRY(16b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(17b,.Lcopy_done)
b .Lcopy_loop_16
ldo -32(len),len
.Lword_loop:
cmpib,COND(>>=),n 3,len,.Lbyte_loop
20: ldw,ma 4(srcspc,src),t1
21: stw,ma t1,4(dstspc,dst)
b .Lword_loop
ldo -4(len),len
ASM_EXCEPTIONTABLE_ENTRY(20b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(21b,.Lcopy_done)
#endif /* CONFIG_64BIT */
/* loop until we are 32-bit aligned */
.Lalign_loop32:
extru dst,31,2,t1
cmpib,=,n 0,t1,.Lcopy_loop_4
20: ldb,ma 1(srcspc,src),t1
21: stb,ma t1,1(dstspc,dst)
b .Lalign_loop32
ldo -1(len),len
ASM_EXCEPTIONTABLE_ENTRY(20b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(21b,.Lcopy_done)
.Lcopy_loop_4:
cmpib,COND(>>=),n 15,len,.Lbyte_loop
10: ldw 0(srcspc,src),t1
11: ldw 4(srcspc,src),t2
12: stw,ma t1,4(dstspc,dst)
13: stw,ma t2,4(dstspc,dst)
14: ldw 8(srcspc,src),t1
15: ldw 12(srcspc,src),t2
ldo 16(src),src
16: stw,ma t1,4(dstspc,dst)
17: stw,ma t2,4(dstspc,dst)
ASM_EXCEPTIONTABLE_ENTRY(10b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(11b,.Lcopy8_fault)
ASM_EXCEPTIONTABLE_ENTRY(12b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(13b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(14b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(15b,.Lcopy8_fault)
ASM_EXCEPTIONTABLE_ENTRY(16b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(17b,.Lcopy_done)
b .Lcopy_loop_4
ldo -16(len),len
.Lbyte_loop:
cmpclr,COND(<>) len,%r0,%r0
b,n .Lcopy_done
20: ldb 0(srcspc,src),t1
ldo 1(src),src
21: stb,ma t1,1(dstspc,dst)
b .Lbyte_loop
ldo -1(len),len
ASM_EXCEPTIONTABLE_ENTRY(20b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(21b,.Lcopy_done)
.Lcopy_done:
bv %r0(%r2)
sub end,dst,ret0
/* src and dst are not aligned the same way. */
/* need to go the hard way */
.Lunaligned_copy:
/* align until dst is 32bit-word-aligned */
extru dst,31,2,t1
cmpib,COND(=),n 0,t1,.Lcopy_dstaligned
20: ldb 0(srcspc,src),t1
ldo 1(src),src
21: stb,ma t1,1(dstspc,dst)
b .Lunaligned_copy
ldo -1(len),len
ASM_EXCEPTIONTABLE_ENTRY(20b,.Lcopy_done)
ASM_EXCEPTIONTABLE_ENTRY(21b,.Lcopy_done)
.Lcopy_dstaligned:
/* store src, dst and len in safe place */
copy src,save_src
copy dst,save_dst
copy len,save_len
/* len now needs give number of words to copy */
SHRREG len,2,len
/*
* Copy from a not-aligned src to an aligned dst using shifts.
* Handles 4 words per loop.
*/
depw,z src,28,2,t0
subi 32,t0,t0
mtsar t0
extru len,31,2,t0
cmpib,= 2,t0,.Lcase2
/* Make src aligned by rounding it down. */
depi 0,31,2,src
cmpiclr,<> 3,t0,%r0
b,n .Lcase3
cmpiclr,<> 1,t0,%r0
b,n .Lcase1
.Lcase0:
cmpb,= %r0,len,.Lcda_finish
nop
1: ldw,ma 4(srcspc,src), a3
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
1: ldw,ma 4(srcspc,src), a0
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
b,n .Ldo3
.Lcase1:
1: ldw,ma 4(srcspc,src), a2
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
1: ldw,ma 4(srcspc,src), a3
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
ldo -1(len),len
cmpb,=,n %r0,len,.Ldo0
.Ldo4:
1: ldw,ma 4(srcspc,src), a0
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
shrpw a2, a3, %sar, t0
1: stw,ma t0, 4(dstspc,dst)
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcopy_done)
.Ldo3:
1: ldw,ma 4(srcspc,src), a1
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
shrpw a3, a0, %sar, t0
1: stw,ma t0, 4(dstspc,dst)
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcopy_done)
.Ldo2:
1: ldw,ma 4(srcspc,src), a2
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
shrpw a0, a1, %sar, t0
1: stw,ma t0, 4(dstspc,dst)
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcopy_done)
.Ldo1:
1: ldw,ma 4(srcspc,src), a3
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
shrpw a1, a2, %sar, t0
1: stw,ma t0, 4(dstspc,dst)
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcopy_done)
ldo -4(len),len
cmpb,<> %r0,len,.Ldo4
nop
.Ldo0:
shrpw a2, a3, %sar, t0
1: stw,ma t0, 4(dstspc,dst)
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcopy_done)
.Lcda_rdfault:
.Lcda_finish:
/* calculate new src, dst and len and jump to byte-copy loop */
sub dst,save_dst,t0
add save_src,t0,src
b .Lbyte_loop
sub save_len,t0,len
.Lcase3:
1: ldw,ma 4(srcspc,src), a0
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
1: ldw,ma 4(srcspc,src), a1
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
b .Ldo2
ldo 1(len),len
.Lcase2:
1: ldw,ma 4(srcspc,src), a1
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
1: ldw,ma 4(srcspc,src), a2
ASM_EXCEPTIONTABLE_ENTRY(1b,.Lcda_rdfault)
b .Ldo1
ldo 2(len),len
/* fault exception fixup handlers: */
#ifdef CONFIG_64BIT
.Lcopy16_fault:
10: b .Lcopy_done
std,ma t1,8(dstspc,dst)
ASM_EXCEPTIONTABLE_ENTRY(10b,.Lcopy_done)
#endif
.Lcopy8_fault:
10: b .Lcopy_done
stw,ma t1,4(dstspc,dst)
ASM_EXCEPTIONTABLE_ENTRY(10b,.Lcopy_done)
.exit
ENDPROC_CFI(pa_memcpy)
.procend
.end
......@@ -2,7 +2,7 @@
* Optimized memory copy routines.
*
* Copyright (C) 2004 Randolph Chung <tausq@debian.org>
* Copyright (C) 2013 Helge Deller <deller@gmx.de>
* Copyright (C) 2013-2017 Helge Deller <deller@gmx.de>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
......@@ -21,474 +21,21 @@
* Portions derived from the GNU C Library
* Copyright (C) 1991, 1997, 2003 Free Software Foundation, Inc.
*
* Several strategies are tried to try to get the best performance for various
* conditions. In the optimal case, we copy 64-bytes in an unrolled loop using
* fp regs. This is followed by loops that copy 32- or 16-bytes at a time using
* general registers. Unaligned copies are handled either by aligning the
* destination and then using shift-and-write method, or in a few cases by
* falling back to a byte-at-a-time copy.
*
* I chose to implement this in C because it is easier to maintain and debug,
* and in my experiments it appears that the C code generated by gcc (3.3/3.4
* at the time of writing) is fairly optimal. Unfortunately some of the
* semantics of the copy routine (exception handling) is difficult to express
* in C, so we have to play some tricks to get it to work.
*
* All the loads and stores are done via explicit asm() code in order to use
* the right space registers.
*
* Testing with various alignments and buffer sizes shows that this code is
* often >10x faster than a simple byte-at-a-time copy, even for strangely
* aligned operands. It is interesting to note that the glibc version
* of memcpy (written in C) is actually quite fast already. This routine is
* able to beat it by 30-40% for aligned copies because of the loop unrolling,
* but in some cases the glibc version is still slightly faster. This lends
* more credibility that gcc can generate very good code as long as we are
* careful.
*
* TODO:
* - cache prefetching needs more experimentation to get optimal settings
* - try not to use the post-increment address modifiers; they create additional
* interlocks
* - replace byte-copy loops with stybs sequences
*/
#ifdef __KERNEL__
#include <linux/module.h>
#include <linux/compiler.h>
#include <linux/uaccess.h>
#define s_space "%%sr1"
#define d_space "%%sr2"
#else
#include "memcpy.h"
#define s_space "%%sr0"
#define d_space "%%sr0"
#define pa_memcpy new2_copy
#endif
DECLARE_PER_CPU(struct exception_data, exception_data);
#define preserve_branch(label) do { \
volatile int dummy = 0; \
/* The following branch is never taken, it's just here to */ \
/* prevent gcc from optimizing away our exception code. */ \
if (unlikely(dummy != dummy)) \
goto label; \
} while (0)
#define get_user_space() (segment_eq(get_fs(), KERNEL_DS) ? 0 : mfsp(3))
#define get_kernel_space() (0)
#define MERGE(w0, sh_1, w1, sh_2) ({ \
unsigned int _r; \
asm volatile ( \
"mtsar %3\n" \
"shrpw %1, %2, %%sar, %0\n" \
: "=r"(_r) \
: "r"(w0), "r"(w1), "r"(sh_2) \
); \
_r; \
})
#define THRESHOLD 16
#ifdef DEBUG_MEMCPY
#define DPRINTF(fmt, args...) do { printk(KERN_DEBUG "%s:%d:%s ", __FILE__, __LINE__, __func__ ); printk(KERN_DEBUG fmt, ##args ); } while (0)
#else
#define DPRINTF(fmt, args...)
#endif
#define def_load_ai_insn(_insn,_sz,_tt,_s,_a,_t,_e) \
__asm__ __volatile__ ( \
"1:\t" #_insn ",ma " #_sz "(" _s ",%1), %0\n\t" \
ASM_EXCEPTIONTABLE_ENTRY(1b,_e) \
: _tt(_t), "+r"(_a) \
: \
: "r8")
#define def_store_ai_insn(_insn,_sz,_tt,_s,_a,_t,_e) \
__asm__ __volatile__ ( \
"1:\t" #_insn ",ma %1, " #_sz "(" _s ",%0)\n\t" \
ASM_EXCEPTIONTABLE_ENTRY(1b,_e) \
: "+r"(_a) \
: _tt(_t) \
: "r8")
#define ldbma(_s, _a, _t, _e) def_load_ai_insn(ldbs,1,"=r",_s,_a,_t,_e)
#define stbma(_s, _t, _a, _e) def_store_ai_insn(stbs,1,"r",_s,_a,_t,_e)
#define ldwma(_s, _a, _t, _e) def_load_ai_insn(ldw,4,"=r",_s,_a,_t,_e)
#define stwma(_s, _t, _a, _e) def_store_ai_insn(stw,4,"r",_s,_a,_t,_e)
#define flddma(_s, _a, _t, _e) def_load_ai_insn(fldd,8,"=f",_s,_a,_t,_e)
#define fstdma(_s, _t, _a, _e) def_store_ai_insn(fstd,8,"f",_s,_a,_t,_e)
#define def_load_insn(_insn,_tt,_s,_o,_a,_t,_e) \
__asm__ __volatile__ ( \
"1:\t" #_insn " " #_o "(" _s ",%1), %0\n\t" \
ASM_EXCEPTIONTABLE_ENTRY(1b,_e) \
: _tt(_t) \
: "r"(_a) \
: "r8")
#define def_store_insn(_insn,_tt,_s,_t,_o,_a,_e) \
__asm__ __volatile__ ( \
"1:\t" #_insn " %0, " #_o "(" _s ",%1)\n\t" \
ASM_EXCEPTIONTABLE_ENTRY(1b,_e) \
: \
: _tt(_t), "r"(_a) \
: "r8")
#define ldw(_s,_o,_a,_t,_e) def_load_insn(ldw,"=r",_s,_o,_a,_t,_e)
#define stw(_s,_t,_o,_a,_e) def_store_insn(stw,"r",_s,_t,_o,_a,_e)
#ifdef CONFIG_PREFETCH
static inline void prefetch_src(const void *addr)
{
__asm__("ldw 0(" s_space ",%0), %%r0" : : "r" (addr));
}
static inline void prefetch_dst(const void *addr)
{
__asm__("ldd 0(" d_space ",%0), %%r0" : : "r" (addr));
}
#else
#define prefetch_src(addr) do { } while(0)
#define prefetch_dst(addr) do { } while(0)
#endif
#define PA_MEMCPY_OK 0
#define PA_MEMCPY_LOAD_ERROR 1
#define PA_MEMCPY_STORE_ERROR 2
/* Copy from a not-aligned src to an aligned dst, using shifts. Handles 4 words
* per loop. This code is derived from glibc.
*/
static noinline unsigned long copy_dstaligned(unsigned long dst,
unsigned long src, unsigned long len)
{
/* gcc complains that a2 and a3 may be uninitialized, but actually
* they cannot be. Initialize a2/a3 to shut gcc up.
*/
register unsigned int a0, a1, a2 = 0, a3 = 0;
int sh_1, sh_2;
/* prefetch_src((const void *)src); */
/* Calculate how to shift a word read at the memory operation
aligned srcp to make it aligned for copy. */
sh_1 = 8 * (src % sizeof(unsigned int));
sh_2 = 8 * sizeof(unsigned int) - sh_1;
/* Make src aligned by rounding it down. */
src &= -sizeof(unsigned int);
switch (len % 4)
{
case 2:
/* a1 = ((unsigned int *) src)[0];
a2 = ((unsigned int *) src)[1]; */
ldw(s_space, 0, src, a1, cda_ldw_exc);
ldw(s_space, 4, src, a2, cda_ldw_exc);
src -= 1 * sizeof(unsigned int);
dst -= 3 * sizeof(unsigned int);
len += 2;
goto do1;
case 3:
/* a0 = ((unsigned int *) src)[0];
a1 = ((unsigned int *) src)[1]; */
ldw(s_space, 0, src, a0, cda_ldw_exc);
ldw(s_space, 4, src, a1, cda_ldw_exc);
src -= 0 * sizeof(unsigned int);
dst -= 2 * sizeof(unsigned int);
len += 1;
goto do2;
case 0:
if (len == 0)
return PA_MEMCPY_OK;
/* a3 = ((unsigned int *) src)[0];
a0 = ((unsigned int *) src)[1]; */
ldw(s_space, 0, src, a3, cda_ldw_exc);
ldw(s_space, 4, src, a0, cda_ldw_exc);
src -=-1 * sizeof(unsigned int);
dst -= 1 * sizeof(unsigned int);
len += 0;
goto do3;
case 1:
/* a2 = ((unsigned int *) src)[0];
a3 = ((unsigned int *) src)[1]; */
ldw(s_space, 0, src, a2, cda_ldw_exc);
ldw(s_space, 4, src, a3, cda_ldw_exc);
src -=-2 * sizeof(unsigned int);
dst -= 0 * sizeof(unsigned int);
len -= 1;
if (len == 0)
goto do0;
goto do4; /* No-op. */
}
do
{
/* prefetch_src((const void *)(src + 4 * sizeof(unsigned int))); */
do4:
/* a0 = ((unsigned int *) src)[0]; */
ldw(s_space, 0, src, a0, cda_ldw_exc);
/* ((unsigned int *) dst)[0] = MERGE (a2, sh_1, a3, sh_2); */
stw(d_space, MERGE (a2, sh_1, a3, sh_2), 0, dst, cda_stw_exc);
do3:
/* a1 = ((unsigned int *) src)[1]; */
ldw(s_space, 4, src, a1, cda_ldw_exc);
/* ((unsigned int *) dst)[1] = MERGE (a3, sh_1, a0, sh_2); */
stw(d_space, MERGE (a3, sh_1, a0, sh_2), 4, dst, cda_stw_exc);
do2:
/* a2 = ((unsigned int *) src)[2]; */
ldw(s_space, 8, src, a2, cda_ldw_exc);
/* ((unsigned int *) dst)[2] = MERGE (a0, sh_1, a1, sh_2); */
stw(d_space, MERGE (a0, sh_1, a1, sh_2), 8, dst, cda_stw_exc);
do1:
/* a3 = ((unsigned int *) src)[3]; */
ldw(s_space, 12, src, a3, cda_ldw_exc);
/* ((unsigned int *) dst)[3] = MERGE (a1, sh_1, a2, sh_2); */
stw(d_space, MERGE (a1, sh_1, a2, sh_2), 12, dst, cda_stw_exc);
src += 4 * sizeof(unsigned int);
dst += 4 * sizeof(unsigned int);
len -= 4;
}
while (len != 0);
do0:
/* ((unsigned int *) dst)[0] = MERGE (a2, sh_1, a3, sh_2); */
stw(d_space, MERGE (a2, sh_1, a3, sh_2), 0, dst, cda_stw_exc);
preserve_branch(handle_load_error);
preserve_branch(handle_store_error);
return PA_MEMCPY_OK;
handle_load_error:
__asm__ __volatile__ ("cda_ldw_exc:\n");
return PA_MEMCPY_LOAD_ERROR;
handle_store_error:
__asm__ __volatile__ ("cda_stw_exc:\n");
return PA_MEMCPY_STORE_ERROR;
}
/* Returns PA_MEMCPY_OK, PA_MEMCPY_LOAD_ERROR or PA_MEMCPY_STORE_ERROR.
* In case of an access fault the faulty address can be read from the per_cpu
* exception data struct. */
static noinline unsigned long pa_memcpy_internal(void *dstp, const void *srcp,
unsigned long len)
{
register unsigned long src, dst, t1, t2, t3;
register unsigned char *pcs, *pcd;
register unsigned int *pws, *pwd;
register double *pds, *pdd;
unsigned long ret;
src = (unsigned long)srcp;
dst = (unsigned long)dstp;
pcs = (unsigned char *)srcp;
pcd = (unsigned char *)dstp;
/* prefetch_src((const void *)srcp); */
if (len < THRESHOLD)
goto byte_copy;
/* Check alignment */
t1 = (src ^ dst);
if (unlikely(t1 & (sizeof(double)-1)))
goto unaligned_copy;
/* src and dst have same alignment. */
/* Copy bytes till we are double-aligned. */
t2 = src & (sizeof(double) - 1);
if (unlikely(t2 != 0)) {
t2 = sizeof(double) - t2;
while (t2 && len) {
/* *pcd++ = *pcs++; */
ldbma(s_space, pcs, t3, pmc_load_exc);
len--;
stbma(d_space, t3, pcd, pmc_store_exc);
t2--;
}
}
pds = (double *)pcs;
pdd = (double *)pcd;
#if 0
/* Copy 8 doubles at a time */
while (len >= 8*sizeof(double)) {
register double r1, r2, r3, r4, r5, r6, r7, r8;
/* prefetch_src((char *)pds + L1_CACHE_BYTES); */
flddma(s_space, pds, r1, pmc_load_exc);
flddma(s_space, pds, r2, pmc_load_exc);
flddma(s_space, pds, r3, pmc_load_exc);
flddma(s_space, pds, r4, pmc_load_exc);
fstdma(d_space, r1, pdd, pmc_store_exc);
fstdma(d_space, r2, pdd, pmc_store_exc);
fstdma(d_space, r3, pdd, pmc_store_exc);
fstdma(d_space, r4, pdd, pmc_store_exc);
#if 0
if (L1_CACHE_BYTES <= 32)
prefetch_src((char *)pds + L1_CACHE_BYTES);
#endif
flddma(s_space, pds, r5, pmc_load_exc);
flddma(s_space, pds, r6, pmc_load_exc);
flddma(s_space, pds, r7, pmc_load_exc);
flddma(s_space, pds, r8, pmc_load_exc);
fstdma(d_space, r5, pdd, pmc_store_exc);
fstdma(d_space, r6, pdd, pmc_store_exc);
fstdma(d_space, r7, pdd, pmc_store_exc);
fstdma(d_space, r8, pdd, pmc_store_exc);
len -= 8*sizeof(double);
}
#endif
pws = (unsigned int *)pds;
pwd = (unsigned int *)pdd;
word_copy:
while (len >= 8*sizeof(unsigned int)) {
register unsigned int r1,r2,r3,r4,r5,r6,r7,r8;
/* prefetch_src((char *)pws + L1_CACHE_BYTES); */
ldwma(s_space, pws, r1, pmc_load_exc);
ldwma(s_space, pws, r2, pmc_load_exc);
ldwma(s_space, pws, r3, pmc_load_exc);
ldwma(s_space, pws, r4, pmc_load_exc);
stwma(d_space, r1, pwd, pmc_store_exc);
stwma(d_space, r2, pwd, pmc_store_exc);
stwma(d_space, r3, pwd, pmc_store_exc);
stwma(d_space, r4, pwd, pmc_store_exc);
ldwma(s_space, pws, r5, pmc_load_exc);
ldwma(s_space, pws, r6, pmc_load_exc);
ldwma(s_space, pws, r7, pmc_load_exc);
ldwma(s_space, pws, r8, pmc_load_exc);
stwma(d_space, r5, pwd, pmc_store_exc);
stwma(d_space, r6, pwd, pmc_store_exc);
stwma(d_space, r7, pwd, pmc_store_exc);
stwma(d_space, r8, pwd, pmc_store_exc);
len -= 8*sizeof(unsigned int);
}
while (len >= 4*sizeof(unsigned int)) {
register unsigned int r1,r2,r3,r4;
ldwma(s_space, pws, r1, pmc_load_exc);
ldwma(s_space, pws, r2, pmc_load_exc);
ldwma(s_space, pws, r3, pmc_load_exc);
ldwma(s_space, pws, r4, pmc_load_exc);
stwma(d_space, r1, pwd, pmc_store_exc);
stwma(d_space, r2, pwd, pmc_store_exc);
stwma(d_space, r3, pwd, pmc_store_exc);
stwma(d_space, r4, pwd, pmc_store_exc);
len -= 4*sizeof(unsigned int);
}
pcs = (unsigned char *)pws;
pcd = (unsigned char *)pwd;
byte_copy:
while (len) {
/* *pcd++ = *pcs++; */
ldbma(s_space, pcs, t3, pmc_load_exc);
stbma(d_space, t3, pcd, pmc_store_exc);
len--;
}
return PA_MEMCPY_OK;
unaligned_copy:
/* possibly we are aligned on a word, but not on a double... */
if (likely((t1 & (sizeof(unsigned int)-1)) == 0)) {
t2 = src & (sizeof(unsigned int) - 1);
if (unlikely(t2 != 0)) {
t2 = sizeof(unsigned int) - t2;
while (t2) {
/* *pcd++ = *pcs++; */
ldbma(s_space, pcs, t3, pmc_load_exc);
stbma(d_space, t3, pcd, pmc_store_exc);
len--;
t2--;
}
}
pws = (unsigned int *)pcs;
pwd = (unsigned int *)pcd;
goto word_copy;
}
/* Align the destination. */
if (unlikely((dst & (sizeof(unsigned int) - 1)) != 0)) {
t2 = sizeof(unsigned int) - (dst & (sizeof(unsigned int) - 1));
while (t2) {
/* *pcd++ = *pcs++; */
ldbma(s_space, pcs, t3, pmc_load_exc);
stbma(d_space, t3, pcd, pmc_store_exc);
len--;
t2--;
}
dst = (unsigned long)pcd;
src = (unsigned long)pcs;
}
ret = copy_dstaligned(dst, src, len / sizeof(unsigned int));
if (ret)
return ret;
pcs += (len & -sizeof(unsigned int));
pcd += (len & -sizeof(unsigned int));
len %= sizeof(unsigned int);
preserve_branch(handle_load_error);
preserve_branch(handle_store_error);
goto byte_copy;
handle_load_error:
__asm__ __volatile__ ("pmc_load_exc:\n");
return PA_MEMCPY_LOAD_ERROR;
handle_store_error:
__asm__ __volatile__ ("pmc_store_exc:\n");
return PA_MEMCPY_STORE_ERROR;
}
/* Returns 0 for success, otherwise, returns number of bytes not transferred. */
static unsigned long pa_memcpy(void *dstp, const void *srcp, unsigned long len)
{
unsigned long ret, fault_addr, reference;
struct exception_data *d;
ret = pa_memcpy_internal(dstp, srcp, len);
if (likely(ret == PA_MEMCPY_OK))
return 0;
/* if a load or store fault occured we can get the faulty addr */
d = this_cpu_ptr(&exception_data);
fault_addr = d->fault_addr;
/* error in load or store? */
if (ret == PA_MEMCPY_LOAD_ERROR)
reference = (unsigned long) srcp;
else
reference = (unsigned long) dstp;
extern unsigned long pa_memcpy(void *dst, const void *src,
unsigned long len);
DPRINTF("pa_memcpy: fault type = %lu, len=%lu fault_addr=%lu ref=%lu\n",
ret, len, fault_addr, reference);
if (fault_addr >= reference)
return len - (fault_addr - reference);
else
return len;
}
#ifdef __KERNEL__
unsigned long __copy_to_user(void __user *dst, const void *src,
unsigned long len)
{
......@@ -537,5 +84,3 @@ long probe_kernel_read(void *dst, const void *src, size_t size)
return __probe_kernel_read(dst, src, size);
}
#endif
......@@ -150,6 +150,23 @@ int fixup_exception(struct pt_regs *regs)
d->fault_space = regs->isr;
d->fault_addr = regs->ior;
/*
* Fix up get_user() and put_user().
* ASM_EXCEPTIONTABLE_ENTRY_EFAULT() sets the least-significant
* bit in the relative address of the fixup routine to indicate
* that %r8 should be loaded with -EFAULT to report a userspace
* access error.
*/
if (fix->fixup & 1) {
regs->gr[8] = -EFAULT;
/* zero target register for get_user() */
if (parisc_acctyp(0, regs->iir) == VM_READ) {
int treg = regs->iir & 0x1f;
regs->gr[treg] = 0;
}
}
regs->iaoq[0] = (unsigned long)&fix->fixup + fix->fixup;
regs->iaoq[0] &= ~3;
/*
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment