2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* linux/arch/arm/kernel/entry-common.S
|
|
|
|
*
|
|
|
|
* Copyright (C) 2000 Russell King
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License version 2 as
|
|
|
|
* published by the Free Software Foundation.
|
|
|
|
*/
|
|
|
|
|
2014-06-30 17:29:12 +02:00
|
|
|
#include <asm/assembler.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
#include <asm/unistd.h>
|
2008-06-21 20:17:27 +02:00
|
|
|
#include <asm/ftrace.h>
|
2009-02-16 11:42:09 +01:00
|
|
|
#include <asm/unwind.h>
|
2017-09-07 17:30:46 +02:00
|
|
|
#include <asm/memory.h>
|
2016-10-18 20:57:01 +02:00
|
|
|
#ifdef CONFIG_AEABI
|
|
|
|
#include <asm/unistd-oabi.h>
|
|
|
|
#endif
|
|
|
|
|
|
|
|
.equ NR_syscalls, __NR_syscalls
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2012-02-07 16:28:22 +01:00
|
|
|
#ifdef CONFIG_NEED_RET_TO_USER
|
|
|
|
#include <mach/entry-macro.S>
|
|
|
|
#else
|
|
|
|
.macro arch_ret_to_user, tmp1, tmp2
|
|
|
|
.endm
|
|
|
|
#endif
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
#include "entry-header.S"
|
|
|
|
|
2017-03-24 17:15:45 +01:00
|
|
|
saved_psr .req r8
|
2017-03-24 17:22:10 +01:00
|
|
|
#if defined(CONFIG_TRACE_IRQFLAGS) || defined(CONFIG_CONTEXT_TRACKING)
|
|
|
|
saved_pc .req r9
|
|
|
|
#define TRACE(x...) x
|
|
|
|
#else
|
2017-03-24 17:15:45 +01:00
|
|
|
saved_pc .req lr
|
2017-03-24 17:22:10 +01:00
|
|
|
#define TRACE(x...)
|
|
|
|
#endif
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2017-11-25 00:54:22 +01:00
|
|
|
.section .entry.text,"ax",%progbits
|
2005-04-17 00:20:36 +02:00
|
|
|
.align 5
|
2018-06-02 14:43:56 +02:00
|
|
|
#if !(IS_ENABLED(CONFIG_TRACE_IRQFLAGS) || IS_ENABLED(CONFIG_CONTEXT_TRACKING) || \
|
|
|
|
IS_ENABLED(CONFIG_DEBUG_RSEQ))
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
2015-08-20 17:13:37 +02:00
|
|
|
* This is the fast syscall return path. We do as little as possible here,
|
|
|
|
* such as avoiding writing r0 to the stack. We only use this path if we
|
2018-06-02 14:43:56 +02:00
|
|
|
* have tracing, context tracking and rseq debug disabled - the overheads
|
|
|
|
* from those features make this path too inefficient.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
|
|
|
ret_fast_syscall:
|
ARM: 8781/1: Fix Thumb-2 syscall return for binutils 2.29+
When building the kernel as Thumb-2 with binutils 2.29 or newer, if the
assembler has seen the .type directive (via ENDPROC()) for a symbol, it
automatically handles the setting of the lowest bit when the symbol is
used with ADR. The badr macro on the other hand handles this lowest bit
manually. This leads to a jump to a wrong address in the wrong state
in the syscall return path:
Internal error: Oops - undefined instruction: 0 [#2] SMP THUMB2
Modules linked in:
CPU: 0 PID: 652 Comm: modprobe Tainted: G D 4.18.0-rc3+ #8
PC is at ret_fast_syscall+0x4/0x62
LR is at sys_brk+0x109/0x128
pc : [<80101004>] lr : [<801c8a35>] psr: 60000013
Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment none
Control: 50c5387d Table: 9e82006a DAC: 00000051
Process modprobe (pid: 652, stack limit = 0x(ptrval))
80101000 <ret_fast_syscall>:
80101000: b672 cpsid i
80101002: f8d9 2008 ldr.w r2, [r9, #8]
80101006: f1b2 4ffe cmp.w r2, #2130706432 ; 0x7f000000
80101184 <local_restart>:
80101184: f8d9 a000 ldr.w sl, [r9]
80101188: e92d 0030 stmdb sp!, {r4, r5}
8010118c: f01a 0ff0 tst.w sl, #240 ; 0xf0
80101190: d117 bne.n 801011c2 <__sys_trace>
80101192: 46ba mov sl, r7
80101194: f5ba 7fc8 cmp.w sl, #400 ; 0x190
80101198: bf28 it cs
8010119a: f04f 0a00 movcs.w sl, #0
8010119e: f3af 8014 nop.w {20}
801011a2: f2af 1ea2 subw lr, pc, #418 ; 0x1a2
To fix this, add a new symbol name which doesn't have ENDPROC used on it
and use that with badr. We can't remove the badr usage since that would
would cause breakage with older binutils.
Signed-off-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
2018-07-13 12:12:22 +02:00
|
|
|
__ret_fast_syscall:
|
2009-02-16 11:42:09 +01:00
|
|
|
UNWIND(.fnstart )
|
|
|
|
UNWIND(.cantunwind )
|
2015-08-20 17:13:37 +02:00
|
|
|
disable_irq_notrace @ disable interrupts
|
2017-09-07 17:30:46 +02:00
|
|
|
ldr r2, [tsk, #TI_ADDR_LIMIT]
|
|
|
|
cmp r2, #TASK_SIZE
|
|
|
|
blne addr_limit_check_failed
|
2015-05-15 12:02:23 +02:00
|
|
|
ldr r1, [tsk, #TI_FLAGS] @ re-check for syscall tracing
|
2017-09-07 17:30:45 +02:00
|
|
|
tst r1, #_TIF_SYSCALL_WORK | _TIF_WORK_MASK
|
2005-04-17 00:20:36 +02:00
|
|
|
bne fast_work_pending
|
2005-04-26 16:20:34 +02:00
|
|
|
|
2017-09-07 17:30:46 +02:00
|
|
|
|
2007-02-16 22:16:32 +01:00
|
|
|
/* perform architecture specific actions before user return */
|
|
|
|
arch_ret_to_user r1, lr
|
|
|
|
|
2009-07-24 13:32:54 +02:00
|
|
|
restore_user_regs fast = 1, offset = S_OFF
|
2009-02-16 11:42:09 +01:00
|
|
|
UNWIND(.fnend )
|
2015-08-20 17:13:37 +02:00
|
|
|
ENDPROC(ret_fast_syscall)
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2015-08-20 17:13:37 +02:00
|
|
|
/* Ok, we need to do extra processing, enter the slow path. */
|
2005-04-17 00:20:36 +02:00
|
|
|
fast_work_pending:
|
|
|
|
str r0, [sp, #S_R0+S_OFF]! @ returned r0
|
2015-08-20 17:13:37 +02:00
|
|
|
/* fall through to work_pending */
|
|
|
|
#else
|
|
|
|
/*
|
2018-06-02 14:43:56 +02:00
|
|
|
* The "replacement" ret_fast_syscall for when tracing, context tracking,
|
|
|
|
* or rseq debug is enabled. As we will need to call out to some C functions,
|
|
|
|
* we save r0 first to avoid needing to save registers around each C function
|
|
|
|
* call.
|
2015-08-20 17:13:37 +02:00
|
|
|
*/
|
|
|
|
ret_fast_syscall:
|
ARM: 8781/1: Fix Thumb-2 syscall return for binutils 2.29+
When building the kernel as Thumb-2 with binutils 2.29 or newer, if the
assembler has seen the .type directive (via ENDPROC()) for a symbol, it
automatically handles the setting of the lowest bit when the symbol is
used with ADR. The badr macro on the other hand handles this lowest bit
manually. This leads to a jump to a wrong address in the wrong state
in the syscall return path:
Internal error: Oops - undefined instruction: 0 [#2] SMP THUMB2
Modules linked in:
CPU: 0 PID: 652 Comm: modprobe Tainted: G D 4.18.0-rc3+ #8
PC is at ret_fast_syscall+0x4/0x62
LR is at sys_brk+0x109/0x128
pc : [<80101004>] lr : [<801c8a35>] psr: 60000013
Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment none
Control: 50c5387d Table: 9e82006a DAC: 00000051
Process modprobe (pid: 652, stack limit = 0x(ptrval))
80101000 <ret_fast_syscall>:
80101000: b672 cpsid i
80101002: f8d9 2008 ldr.w r2, [r9, #8]
80101006: f1b2 4ffe cmp.w r2, #2130706432 ; 0x7f000000
80101184 <local_restart>:
80101184: f8d9 a000 ldr.w sl, [r9]
80101188: e92d 0030 stmdb sp!, {r4, r5}
8010118c: f01a 0ff0 tst.w sl, #240 ; 0xf0
80101190: d117 bne.n 801011c2 <__sys_trace>
80101192: 46ba mov sl, r7
80101194: f5ba 7fc8 cmp.w sl, #400 ; 0x190
80101198: bf28 it cs
8010119a: f04f 0a00 movcs.w sl, #0
8010119e: f3af 8014 nop.w {20}
801011a2: f2af 1ea2 subw lr, pc, #418 ; 0x1a2
To fix this, add a new symbol name which doesn't have ENDPROC used on it
and use that with badr. We can't remove the badr usage since that would
would cause breakage with older binutils.
Signed-off-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
2018-07-13 12:12:22 +02:00
|
|
|
__ret_fast_syscall:
|
2015-08-20 17:13:37 +02:00
|
|
|
UNWIND(.fnstart )
|
|
|
|
UNWIND(.cantunwind )
|
|
|
|
str r0, [sp, #S_R0 + S_OFF]! @ save returned r0
|
2018-06-02 14:43:56 +02:00
|
|
|
#if IS_ENABLED(CONFIG_DEBUG_RSEQ)
|
|
|
|
/* do_rseq_syscall needs interrupts enabled. */
|
|
|
|
mov r0, sp @ 'regs'
|
|
|
|
bl do_rseq_syscall
|
|
|
|
#endif
|
2015-08-20 17:13:37 +02:00
|
|
|
disable_irq_notrace @ disable interrupts
|
2017-09-07 17:30:46 +02:00
|
|
|
ldr r2, [tsk, #TI_ADDR_LIMIT]
|
|
|
|
cmp r2, #TASK_SIZE
|
|
|
|
blne addr_limit_check_failed
|
2015-08-20 17:13:37 +02:00
|
|
|
ldr r1, [tsk, #TI_FLAGS] @ re-check for syscall tracing
|
2017-09-07 17:30:45 +02:00
|
|
|
tst r1, #_TIF_SYSCALL_WORK | _TIF_WORK_MASK
|
2015-08-20 17:13:37 +02:00
|
|
|
beq no_work_pending
|
|
|
|
UNWIND(.fnend )
|
|
|
|
ENDPROC(ret_fast_syscall)
|
|
|
|
|
|
|
|
/* Slower path - fall through to work_pending */
|
|
|
|
#endif
|
|
|
|
|
|
|
|
tst r1, #_TIF_SYSCALL_WORK
|
|
|
|
bne __sys_trace_return_nosave
|
|
|
|
slow_work_pending:
|
2005-04-17 00:20:36 +02:00
|
|
|
mov r0, sp @ 'regs'
|
|
|
|
mov r2, why @ 'syscall'
|
2012-07-19 18:47:55 +02:00
|
|
|
bl do_work_pending
|
2012-07-19 18:48:50 +02:00
|
|
|
cmp r0, #0
|
2012-07-19 18:48:21 +02:00
|
|
|
beq no_work_pending
|
2012-07-19 18:48:50 +02:00
|
|
|
movlt scno, #(__NR_restart_syscall - __NR_SYSCALL_BASE)
|
2012-07-19 18:48:21 +02:00
|
|
|
ldmia sp, {r0 - r6} @ have to reload r0 - r6
|
|
|
|
b local_restart @ ... and off we go
|
2015-08-06 19:50:27 +02:00
|
|
|
ENDPROC(ret_fast_syscall)
|
2012-07-19 18:48:21 +02:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* "slow" syscall return path. "why" tells us if this was a real syscall.
|
2015-08-20 17:13:37 +02:00
|
|
|
* IRQs may be enabled here, so always disable them. Note that we use the
|
|
|
|
* "notrace" version to avoid calling into the tracing code unnecessarily.
|
|
|
|
* do_work_pending() will update this state if necessary.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
|
|
|
ENTRY(ret_to_user)
|
|
|
|
ret_slow_syscall:
|
2018-06-02 14:43:56 +02:00
|
|
|
#if IS_ENABLED(CONFIG_DEBUG_RSEQ)
|
|
|
|
/* do_rseq_syscall needs interrupts enabled. */
|
|
|
|
enable_irq_notrace @ enable interrupts
|
|
|
|
mov r0, sp @ 'regs'
|
|
|
|
bl do_rseq_syscall
|
|
|
|
#endif
|
2015-08-20 17:13:37 +02:00
|
|
|
disable_irq_notrace @ disable interrupts
|
2011-06-05 03:24:58 +02:00
|
|
|
ENTRY(ret_to_user_from_irq)
|
2017-09-07 17:30:46 +02:00
|
|
|
ldr r2, [tsk, #TI_ADDR_LIMIT]
|
|
|
|
cmp r2, #TASK_SIZE
|
|
|
|
blne addr_limit_check_failed
|
2005-04-17 00:20:36 +02:00
|
|
|
ldr r1, [tsk, #TI_FLAGS]
|
|
|
|
tst r1, #_TIF_WORK_MASK
|
2015-08-20 17:13:37 +02:00
|
|
|
bne slow_work_pending
|
2005-04-17 00:20:36 +02:00
|
|
|
no_work_pending:
|
2015-08-20 17:13:37 +02:00
|
|
|
asm_trace_hardirqs_on save = 0
|
2013-03-28 12:44:25 +01:00
|
|
|
|
2007-02-16 22:16:32 +01:00
|
|
|
/* perform architecture specific actions before user return */
|
|
|
|
arch_ret_to_user r1, lr
|
2013-03-28 22:54:40 +01:00
|
|
|
ct_user_enter save = 0
|
2007-02-16 22:16:32 +01:00
|
|
|
|
2009-07-24 13:32:54 +02:00
|
|
|
restore_user_regs fast = 0, offset = 0
|
2011-06-05 03:24:58 +02:00
|
|
|
ENDPROC(ret_to_user_from_irq)
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(ret_to_user)
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* This is how we return from a fork.
|
|
|
|
*/
|
|
|
|
ENTRY(ret_from_fork)
|
|
|
|
bl schedule_tail
|
2012-10-11 04:23:29 +02:00
|
|
|
cmp r5, #0
|
|
|
|
movne r0, r4
|
2015-04-21 15:17:25 +02:00
|
|
|
badrne lr, 1f
|
2014-06-30 17:29:12 +02:00
|
|
|
retne r5
|
ARM: fix oops on initial entry to userspace with Thumb2 kernels
Daniel Mack reports an oops at boot with the latest kernels:
Internal error: Oops - undefined instruction: 0 [#1] SMP THUMB2
Modules linked in:
CPU: 0 Not tainted (3.6.0-11057-g584df1d #145)
PC is at cpsw_probe+0x45a/0x9ac
LR is at trace_hardirqs_on_caller+0x8f/0xfc
pc : [<c03493de>] lr : [<c005e81f>] psr: 60000113
sp : cf055fb0 ip : 00000000 fp : 00000000
r10: 00000000 r9 : 00000000 r8 : 00000000
r7 : 00000000 r6 : 00000000 r5 : c0344555 r4 : 00000000
r3 : cf057a40 r2 : 00000000 r1 : 00000001 r0 : 00000000
Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user
Control: 50c5387d Table: 8f3f4019 DAC: 00000015
Process init (pid: 1, stack limit = 0xcf054240)
Stack: (0xcf055fb0 to 0xcf056000)
5fa0: 00000001 00000000 00000000 00000000
5fc0: cf055fb0 c000d1a8 00000000 00000000 00000000 00000000 00000000 00000000
5fe0: 00000000 be9b3f10 00000000 b6f6add0 00000010 00000000 aaaabfaf a8babbaa
The analysis of this is as follows. In init/main.c, we issue:
kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
This creates a new thread, which falls through to the ret_from_fork
assembly, with r4 set NULL and r5 set to kernel_init. You can see
this in your oops dump register set - r5 is 0xc0344555, which is the
address of kernel_init plus 1 which marks the function as Thumb code.
Now, let's look at this code a little closer - this is what the
disassembly looks like:
c000d180 <ret_from_fork>:
c000d180: f03a fe08 bl c0047d94 <schedule_tail>
c000d184: 2d00 cmp r5, #0
c000d186: bf1e ittt ne
c000d188: 4620 movne r0, r4
c000d18a: 46fe movne lr, pc <-- XXXXXXX
c000d18c: 46af movne pc, r5
c000d18e: 46e9 mov r9, sp
c000d190: ea4f 3959 mov.w r9, r9, lsr #13
c000d194: ea4f 3949 mov.w r9, r9, lsl #13
c000d198: e7c8 b.n c000d12c <ret_to_user>
c000d19a: bf00 nop
c000d19c: f3af 8000 nop.w
This code was introduced in 9fff2fa0db911 (arm: switch to saner
kernel_execve() semantics). I have marked one instruction, and it's
the significant one - I'll come back to that later.
Eventually, having had a successful call to kernel_execve(), kernel_init()
returns zero.
In returning, it uses the value in 'lr' which was set by the instruction
I marked above. Unfortunately, this causes lr to contain 0xc000d18e -
an even address. This switches the ISA to ARM on return but with a non
word aligned PC value.
So, what do we end up executing? Well, not the instructions above - yes
the opcodes, but they don't mean the same thing in ARM mode. In ARM mode,
it looks like this instead:
c000d18c: 46e946af strbtmi r4, [r9], pc, lsr #13
c000d190: 3959ea4f ldmdbcc r9, {r0, r1, r2, r3, r6, r9, fp, sp, lr, pc}^
c000d194: 3949ea4f stmdbcc r9, {r0, r1, r2, r3, r6, r9, fp, sp, lr, pc}^
c000d198: bf00e7c8 svclt 0x0000e7c8
c000d19c: 8000f3af andhi pc, r0, pc, lsr #7
c000d1a0: e88db092 stm sp, {r1, r4, r7, ip, sp, pc}
c000d1a4: 46e81fff ; <UNDEFINED> instruction: 0x46e81fff
c000d1a8: 8a00f3ef bhi 0xc004a16c
c000d1ac: 0a0cf08a beq 0xc03493dc
I have included more above, because it's relevant. The PSR flags which
we can see in the oops dump are nZCv, so Z and C are set.
All the above ARM instructions are not executed, except for two.
c000d1a0, which has no writeback, and writes below the current stack
pointer (and that data is lost when we take the next exception.) The
other instruction which is executed is c000d1ac, which takes us to...
0xc03493dc. However, remember that bit 1 of the PC got set. So that
makes the PC value 0xc03493de.
And that value is the value we find in the oops dump for PC. What is
the instruction here when interpreted in ARM mode?
0: f71e150c ; <UNDEFINED> instruction: 0xf71e150c
and there we have our undefined instruction (remember that the 'never'
condition code, 0xf, has been deprecated and is now always executed as
it is now being used for additional instructions.)
This path also nicely explains the state of the stack we see in the oops
dump too.
The above is a consistent and sane story for how we got to the oops
dump, which all stems from the instruction at 0xc000d18a being wrong.
Reported-by: Daniel Mack <zonque@gmail.com>
Tested-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-15 01:16:49 +02:00
|
|
|
1: get_thread_info tsk
|
2005-04-17 00:20:36 +02:00
|
|
|
b ret_slow_syscall
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(ret_from_fork)
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*=============================================================================
|
|
|
|
* SWI handler
|
|
|
|
*-----------------------------------------------------------------------------
|
|
|
|
*/
|
|
|
|
|
|
|
|
.align 5
|
|
|
|
ENTRY(vector_swi)
|
2010-05-21 19:06:42 +02:00
|
|
|
#ifdef CONFIG_CPU_V7M
|
|
|
|
v7m_exception_entry
|
|
|
|
#else
|
2016-05-10 17:34:27 +02:00
|
|
|
sub sp, sp, #PT_REGS_SIZE
|
2005-04-26 16:20:34 +02:00
|
|
|
stmia sp, {r0 - r12} @ Calling r0 - r12
|
2009-07-24 13:32:54 +02:00
|
|
|
ARM( add r8, sp, #S_PC )
|
|
|
|
ARM( stmdb r8, {sp, lr}^ ) @ Calling sp, lr
|
|
|
|
THUMB( mov r8, sp )
|
|
|
|
THUMB( store_user_sp_lr r8, r10, S_SP ) @ calling sp, lr
|
2017-03-24 17:15:45 +01:00
|
|
|
mrs saved_psr, spsr @ called from non-FIQ mode, so ok.
|
2017-03-24 17:22:10 +01:00
|
|
|
TRACE( mov saved_pc, lr )
|
2017-03-24 17:15:45 +01:00
|
|
|
str saved_pc, [sp, #S_PC] @ Save calling PC
|
|
|
|
str saved_psr, [sp, #S_PSR] @ Save CPSR
|
2005-04-26 16:20:34 +02:00
|
|
|
str r0, [sp, #S_OLD_R0] @ Save OLD_R0
|
2010-05-21 19:06:42 +02:00
|
|
|
#endif
|
2005-04-17 00:20:36 +02:00
|
|
|
zero_fp
|
2014-08-28 14:08:14 +02:00
|
|
|
alignment_trap r10, ip, __cr_alignment
|
2017-03-24 17:23:25 +01:00
|
|
|
asm_trace_hardirqs_on save=0
|
|
|
|
enable_irq_notrace
|
|
|
|
ct_user_exit save=0
|
2013-06-05 12:25:13 +02:00
|
|
|
|
2005-04-26 16:19:24 +02:00
|
|
|
/*
|
|
|
|
* Get the system call number.
|
|
|
|
*/
|
2006-01-14 17:31:29 +01:00
|
|
|
|
2006-01-14 17:36:12 +01:00
|
|
|
#if defined(CONFIG_OABI_COMPAT)
|
2006-01-14 17:31:29 +01:00
|
|
|
|
2006-01-14 17:36:12 +01:00
|
|
|
/*
|
|
|
|
* If we have CONFIG_OABI_COMPAT then we need to look at the swi
|
|
|
|
* value to determine if it is an EABI or an old ABI call.
|
|
|
|
*/
|
|
|
|
#ifdef CONFIG_ARM_THUMB
|
2017-03-24 17:15:45 +01:00
|
|
|
tst saved_psr, #PSR_T_BIT
|
2006-01-14 17:36:12 +01:00
|
|
|
movne r10, #0 @ no thumb OABI emulation
|
2017-03-24 17:15:45 +01:00
|
|
|
USER( ldreq r10, [saved_pc, #-4] ) @ get SWI instruction
|
2006-01-14 17:36:12 +01:00
|
|
|
#else
|
2017-03-24 17:15:45 +01:00
|
|
|
USER( ldr r10, [saved_pc, #-4] ) @ get SWI instruction
|
2006-01-14 17:36:12 +01:00
|
|
|
#endif
|
2013-02-12 19:59:57 +01:00
|
|
|
ARM_BE8(rev r10, r10) @ little endian instruction
|
2006-01-14 17:36:12 +01:00
|
|
|
|
|
|
|
#elif defined(CONFIG_AEABI)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Pure EABI user space always put syscall number into scno (r7).
|
|
|
|
*/
|
2006-01-14 17:31:29 +01:00
|
|
|
#elif defined(CONFIG_ARM_THUMB)
|
2006-01-14 17:36:12 +01:00
|
|
|
/* Legacy ABI only, possibly thumb mode. */
|
2017-03-24 17:15:45 +01:00
|
|
|
tst saved_psr, #PSR_T_BIT @ this is SPSR from save_user_regs
|
2005-04-26 16:19:24 +02:00
|
|
|
addne scno, r7, #__NR_SYSCALL_BASE @ put OS number in
|
2017-03-24 17:15:45 +01:00
|
|
|
USER( ldreq scno, [saved_pc, #-4] )
|
2006-01-14 17:36:12 +01:00
|
|
|
|
2005-04-26 16:19:24 +02:00
|
|
|
#else
|
2006-01-14 17:36:12 +01:00
|
|
|
/* Legacy ABI only. */
|
2017-03-24 17:15:45 +01:00
|
|
|
USER( ldr scno, [saved_pc, #-4] ) @ get SWI instruction
|
2005-04-26 16:19:24 +02:00
|
|
|
#endif
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2017-03-24 17:15:45 +01:00
|
|
|
/* saved_psr and saved_pc are now dead */
|
|
|
|
|
2015-08-20 11:32:02 +02:00
|
|
|
uaccess_disable tbl
|
|
|
|
|
2006-01-14 17:36:12 +01:00
|
|
|
adr tbl, sys_call_table @ load syscall table pointer
|
|
|
|
|
|
|
|
#if defined(CONFIG_OABI_COMPAT)
|
|
|
|
/*
|
|
|
|
* If the swi argument is zero, this is an EABI call and we do nothing.
|
|
|
|
*
|
|
|
|
* If this is an old ABI call, get the syscall number into scno and
|
|
|
|
* get the old ABI syscall table address.
|
|
|
|
*/
|
|
|
|
bics r10, r10, #0xff000000
|
|
|
|
eorne scno, r10, #__NR_OABI_SYSCALL_BASE
|
|
|
|
ldrne tbl, =sys_oabi_call_table
|
|
|
|
#elif !defined(CONFIG_AEABI)
|
2005-04-17 00:20:36 +02:00
|
|
|
bic scno, scno, #0xff000000 @ mask off SWI op-code
|
2005-04-26 16:19:24 +02:00
|
|
|
eor scno, scno, #__NR_SYSCALL_BASE @ check OS number
|
2006-01-14 17:31:29 +01:00
|
|
|
#endif
|
2017-03-24 17:19:12 +01:00
|
|
|
get_thread_info tsk
|
2017-03-24 17:23:25 +01:00
|
|
|
/*
|
|
|
|
* Reload the registers that may have been corrupted on entry to
|
|
|
|
* the syscall assembly (by tracing or context tracking.)
|
|
|
|
*/
|
|
|
|
TRACE( ldmia sp, {r0 - r3} )
|
2006-01-14 17:36:12 +01:00
|
|
|
|
2012-07-19 18:48:21 +02:00
|
|
|
local_restart:
|
2010-08-27 00:08:35 +02:00
|
|
|
ldr r10, [tsk, #TI_FLAGS] @ check for syscall tracing
|
2006-01-14 17:31:29 +01:00
|
|
|
stmdb sp!, {r4, r5} @ push fifth and sixth args
|
2010-08-27 00:08:35 +02:00
|
|
|
|
2012-01-03 20:23:09 +01:00
|
|
|
tst r10, #_TIF_SYSCALL_WORK @ are we tracing syscalls?
|
2005-04-17 00:20:36 +02:00
|
|
|
bne __sys_trace
|
|
|
|
|
ARM: 8781/1: Fix Thumb-2 syscall return for binutils 2.29+
When building the kernel as Thumb-2 with binutils 2.29 or newer, if the
assembler has seen the .type directive (via ENDPROC()) for a symbol, it
automatically handles the setting of the lowest bit when the symbol is
used with ADR. The badr macro on the other hand handles this lowest bit
manually. This leads to a jump to a wrong address in the wrong state
in the syscall return path:
Internal error: Oops - undefined instruction: 0 [#2] SMP THUMB2
Modules linked in:
CPU: 0 PID: 652 Comm: modprobe Tainted: G D 4.18.0-rc3+ #8
PC is at ret_fast_syscall+0x4/0x62
LR is at sys_brk+0x109/0x128
pc : [<80101004>] lr : [<801c8a35>] psr: 60000013
Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment none
Control: 50c5387d Table: 9e82006a DAC: 00000051
Process modprobe (pid: 652, stack limit = 0x(ptrval))
80101000 <ret_fast_syscall>:
80101000: b672 cpsid i
80101002: f8d9 2008 ldr.w r2, [r9, #8]
80101006: f1b2 4ffe cmp.w r2, #2130706432 ; 0x7f000000
80101184 <local_restart>:
80101184: f8d9 a000 ldr.w sl, [r9]
80101188: e92d 0030 stmdb sp!, {r4, r5}
8010118c: f01a 0ff0 tst.w sl, #240 ; 0xf0
80101190: d117 bne.n 801011c2 <__sys_trace>
80101192: 46ba mov sl, r7
80101194: f5ba 7fc8 cmp.w sl, #400 ; 0x190
80101198: bf28 it cs
8010119a: f04f 0a00 movcs.w sl, #0
8010119e: f3af 8014 nop.w {20}
801011a2: f2af 1ea2 subw lr, pc, #418 ; 0x1a2
To fix this, add a new symbol name which doesn't have ENDPROC used on it
and use that with badr. We can't remove the badr usage since that would
would cause breakage with older binutils.
Signed-off-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
2018-07-13 12:12:22 +02:00
|
|
|
invoke_syscall tbl, scno, r10, __ret_fast_syscall
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
add r1, sp, #S_OFF
|
2013-09-19 11:32:20 +02:00
|
|
|
2: cmp scno, #(__ARM_NR_BASE - __NR_SYSCALL_BASE)
|
2005-04-26 16:19:24 +02:00
|
|
|
eor r0, scno, #__NR_SYSCALL_BASE @ put OS number back
|
2013-05-13 20:16:34 +02:00
|
|
|
bcs arm_syscall
|
2013-09-19 11:32:20 +02:00
|
|
|
mov why, #0 @ no longer a real syscall
|
2005-04-17 00:20:36 +02:00
|
|
|
b sys_ni_syscall @ not private func
|
2013-06-05 12:25:13 +02:00
|
|
|
|
|
|
|
#if defined(CONFIG_OABI_COMPAT) || !defined(CONFIG_AEABI)
|
|
|
|
/*
|
|
|
|
* We failed to handle a fault trying to access the page
|
|
|
|
* containing the swi instruction, but we're not really in a
|
|
|
|
* position to return -EFAULT. Instead, return back to the
|
|
|
|
* instruction and re-enter the user fault handling path trying
|
|
|
|
* to page it in. This will likely result in sending SEGV to the
|
|
|
|
* current task.
|
|
|
|
*/
|
|
|
|
9001:
|
2017-03-24 17:15:45 +01:00
|
|
|
sub lr, saved_pc, #4
|
2013-06-05 12:25:13 +02:00
|
|
|
str lr, [sp, #S_PC]
|
2017-03-24 17:19:12 +01:00
|
|
|
get_thread_info tsk
|
2013-06-05 12:25:13 +02:00
|
|
|
b ret_fast_syscall
|
|
|
|
#endif
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(vector_swi)
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* This is the really slow path. We're going to be doing
|
|
|
|
* context switches, and waiting for our parent to respond.
|
|
|
|
*/
|
|
|
|
__sys_trace:
|
2012-07-06 16:50:14 +02:00
|
|
|
mov r1, scno
|
|
|
|
add r0, sp, #S_OFF
|
|
|
|
bl syscall_trace_enter
|
2018-05-11 12:16:22 +02:00
|
|
|
mov scno, r0
|
|
|
|
invoke_syscall tbl, scno, r10, __sys_trace_return, reload=1
|
2012-11-15 22:12:17 +01:00
|
|
|
cmp scno, #-1 @ skip the syscall?
|
|
|
|
bne 2b
|
|
|
|
add sp, sp, #S_OFF @ restore stack
|
|
|
|
b ret_slow_syscall
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
__sys_trace_return:
|
|
|
|
str r0, [sp, #S_R0 + S_OFF]! @ save returned r0
|
2012-07-06 16:50:14 +02:00
|
|
|
mov r0, sp
|
|
|
|
bl syscall_trace_exit
|
2005-04-17 00:20:36 +02:00
|
|
|
b ret_slow_syscall
|
|
|
|
|
2015-08-20 17:13:37 +02:00
|
|
|
__sys_trace_return_nosave:
|
2015-08-20 18:39:32 +02:00
|
|
|
enable_irq_notrace
|
2015-08-20 17:13:37 +02:00
|
|
|
mov r0, sp
|
|
|
|
bl syscall_trace_exit
|
|
|
|
b ret_slow_syscall
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
.align 5
|
|
|
|
#ifdef CONFIG_ALIGNMENT_TRAP
|
|
|
|
.type __cr_alignment, #object
|
|
|
|
__cr_alignment:
|
|
|
|
.word cr_alignment
|
2006-01-14 17:36:12 +01:00
|
|
|
#endif
|
|
|
|
.ltorg
|
|
|
|
|
2016-10-18 20:57:01 +02:00
|
|
|
.macro syscall_table_start, sym
|
|
|
|
.equ __sys_nr, 0
|
|
|
|
.type \sym, #object
|
|
|
|
ENTRY(\sym)
|
|
|
|
.endm
|
|
|
|
|
|
|
|
.macro syscall, nr, func
|
|
|
|
.ifgt __sys_nr - \nr
|
|
|
|
.error "Duplicated/unorded system call entry"
|
|
|
|
.endif
|
|
|
|
.rept \nr - __sys_nr
|
|
|
|
.long sys_ni_syscall
|
|
|
|
.endr
|
|
|
|
.long \func
|
|
|
|
.equ __sys_nr, \nr + 1
|
|
|
|
.endm
|
|
|
|
|
|
|
|
.macro syscall_table_end, sym
|
|
|
|
.ifgt __sys_nr - __NR_syscalls
|
|
|
|
.error "System call table too big"
|
|
|
|
.endif
|
|
|
|
.rept __NR_syscalls - __sys_nr
|
|
|
|
.long sys_ni_syscall
|
|
|
|
.endr
|
|
|
|
.size \sym, . - \sym
|
|
|
|
.endm
|
|
|
|
|
|
|
|
#define NATIVE(nr, func) syscall nr, func
|
|
|
|
|
2006-01-14 17:36:12 +01:00
|
|
|
/*
|
|
|
|
* This is the syscall table declaration for native ABI syscalls.
|
|
|
|
* With EABI a couple syscalls are obsolete and defined as sys_ni_syscall.
|
|
|
|
*/
|
2016-10-18 20:57:01 +02:00
|
|
|
syscall_table_start sys_call_table
|
|
|
|
#define COMPAT(nr, native, compat) syscall nr, native
|
2006-01-14 17:36:12 +01:00
|
|
|
#ifdef CONFIG_AEABI
|
2016-10-18 20:57:01 +02:00
|
|
|
#include <calls-eabi.S>
|
2006-01-14 17:36:12 +01:00
|
|
|
#else
|
2016-10-18 20:57:01 +02:00
|
|
|
#include <calls-oabi.S>
|
2005-04-17 00:20:36 +02:00
|
|
|
#endif
|
2016-10-18 20:57:01 +02:00
|
|
|
#undef COMPAT
|
|
|
|
syscall_table_end sys_call_table
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*============================================================================
|
|
|
|
* Special system call wrappers
|
|
|
|
*/
|
|
|
|
@ r0 = syscall number
|
2005-12-17 16:25:42 +01:00
|
|
|
@ r8 = syscall table
|
2005-04-17 00:20:36 +02:00
|
|
|
sys_syscall:
|
2006-05-16 15:25:55 +02:00
|
|
|
bic scno, r0, #__NR_OABI_SYSCALL_BASE
|
2005-04-17 00:20:36 +02:00
|
|
|
cmp scno, #__NR_syscall - __NR_SYSCALL_BASE
|
|
|
|
cmpne scno, #NR_syscalls @ check range
|
2018-05-11 12:16:22 +02:00
|
|
|
#ifdef CONFIG_CPU_SPECTRE
|
|
|
|
movhs scno, #0
|
|
|
|
csdb
|
|
|
|
#endif
|
2005-04-17 00:20:36 +02:00
|
|
|
stmloia sp, {r5, r6} @ shuffle args
|
|
|
|
movlo r0, r1
|
|
|
|
movlo r1, r2
|
|
|
|
movlo r2, r3
|
|
|
|
movlo r3, r4
|
|
|
|
ldrlo pc, [tbl, scno, lsl #2]
|
|
|
|
b sys_ni_syscall
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_syscall)
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
sys_sigreturn_wrapper:
|
|
|
|
add r0, sp, #S_OFF
|
arm: fix really nasty sigreturn bug
If a signal hits us outside of a syscall and another gets delivered
when we are in sigreturn (e.g. because it had been in sa_mask for
the first one and got sent to us while we'd been in the first handler),
we have a chance of returning from the second handler to location one
insn prior to where we ought to return. If r0 happens to contain -513
(-ERESTARTNOINTR), sigreturn will get confused into doing restart
syscall song and dance.
Incredible joy to debug, since it manifests as random, infrequent and
very hard to reproduce double execution of instructions in userland
code...
The fix is simple - mark it "don't bother with restarts" in wrapper,
i.e. set r8 to 0 in sys_sigreturn and sys_rt_sigreturn wrappers,
suppressing the syscall restart handling on return from these guys.
They can't legitimately return a restart-worthy error anyway.
Testcase:
#include <unistd.h>
#include <signal.h>
#include <stdlib.h>
#include <sys/time.h>
#include <errno.h>
void f(int n)
{
__asm__ __volatile__(
"ldr r0, [%0]\n"
"b 1f\n"
"b 2f\n"
"1:b .\n"
"2:\n" : : "r"(&n));
}
void handler1(int sig) { }
void handler2(int sig) { raise(1); }
void handler3(int sig) { exit(0); }
main()
{
struct sigaction s = {.sa_handler = handler2};
struct itimerval t1 = { .it_value = {1} };
struct itimerval t2 = { .it_value = {2} };
signal(1, handler1);
sigemptyset(&s.sa_mask);
sigaddset(&s.sa_mask, 1);
sigaction(SIGALRM, &s, NULL);
signal(SIGVTALRM, handler3);
setitimer(ITIMER_REAL, &t1, NULL);
setitimer(ITIMER_VIRTUAL, &t2, NULL);
f(-513); /* -ERESTARTNOINTR */
write(1, "buggered\n", 9);
return 1;
}
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-09-17 15:34:39 +02:00
|
|
|
mov why, #0 @ prevent syscall restart handling
|
2005-04-17 00:20:36 +02:00
|
|
|
b sys_sigreturn
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_sigreturn_wrapper)
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
sys_rt_sigreturn_wrapper:
|
|
|
|
add r0, sp, #S_OFF
|
arm: fix really nasty sigreturn bug
If a signal hits us outside of a syscall and another gets delivered
when we are in sigreturn (e.g. because it had been in sa_mask for
the first one and got sent to us while we'd been in the first handler),
we have a chance of returning from the second handler to location one
insn prior to where we ought to return. If r0 happens to contain -513
(-ERESTARTNOINTR), sigreturn will get confused into doing restart
syscall song and dance.
Incredible joy to debug, since it manifests as random, infrequent and
very hard to reproduce double execution of instructions in userland
code...
The fix is simple - mark it "don't bother with restarts" in wrapper,
i.e. set r8 to 0 in sys_sigreturn and sys_rt_sigreturn wrappers,
suppressing the syscall restart handling on return from these guys.
They can't legitimately return a restart-worthy error anyway.
Testcase:
#include <unistd.h>
#include <signal.h>
#include <stdlib.h>
#include <sys/time.h>
#include <errno.h>
void f(int n)
{
__asm__ __volatile__(
"ldr r0, [%0]\n"
"b 1f\n"
"b 2f\n"
"1:b .\n"
"2:\n" : : "r"(&n));
}
void handler1(int sig) { }
void handler2(int sig) { raise(1); }
void handler3(int sig) { exit(0); }
main()
{
struct sigaction s = {.sa_handler = handler2};
struct itimerval t1 = { .it_value = {1} };
struct itimerval t2 = { .it_value = {2} };
signal(1, handler1);
sigemptyset(&s.sa_mask);
sigaddset(&s.sa_mask, 1);
sigaction(SIGALRM, &s, NULL);
signal(SIGVTALRM, handler3);
setitimer(ITIMER_REAL, &t1, NULL);
setitimer(ITIMER_VIRTUAL, &t2, NULL);
f(-513); /* -ERESTARTNOINTR */
write(1, "buggered\n", 9);
return 1;
}
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-09-17 15:34:39 +02:00
|
|
|
mov why, #0 @ prevent syscall restart handling
|
2005-04-17 00:20:36 +02:00
|
|
|
b sys_rt_sigreturn
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_rt_sigreturn_wrapper)
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2006-01-14 17:35:03 +01:00
|
|
|
sys_statfs64_wrapper:
|
|
|
|
teq r1, #88
|
|
|
|
moveq r1, #84
|
|
|
|
b sys_statfs64
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_statfs64_wrapper)
|
2006-01-14 17:35:03 +01:00
|
|
|
|
|
|
|
sys_fstatfs64_wrapper:
|
|
|
|
teq r1, #88
|
|
|
|
moveq r1, #84
|
|
|
|
b sys_fstatfs64
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_fstatfs64_wrapper)
|
2006-01-14 17:35:03 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* Note: off_4k (r5) is always units of 4K. If we can't do the requested
|
|
|
|
* offset, we return EINVAL.
|
|
|
|
*/
|
|
|
|
sys_mmap2:
|
|
|
|
str r5, [sp, #4]
|
2009-11-30 23:37:04 +01:00
|
|
|
b sys_mmap_pgoff
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_mmap2)
|
2006-01-14 17:35:31 +01:00
|
|
|
|
|
|
|
#ifdef CONFIG_OABI_COMPAT
|
2006-01-14 17:36:12 +01:00
|
|
|
|
2006-01-14 17:35:31 +01:00
|
|
|
/*
|
|
|
|
* These are syscalls with argument register differences
|
|
|
|
*/
|
|
|
|
|
|
|
|
sys_oabi_pread64:
|
|
|
|
stmia sp, {r3, r4}
|
|
|
|
b sys_pread64
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_oabi_pread64)
|
2006-01-14 17:35:31 +01:00
|
|
|
|
|
|
|
sys_oabi_pwrite64:
|
|
|
|
stmia sp, {r3, r4}
|
|
|
|
b sys_pwrite64
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_oabi_pwrite64)
|
2006-01-14 17:35:31 +01:00
|
|
|
|
|
|
|
sys_oabi_truncate64:
|
|
|
|
mov r3, r2
|
|
|
|
mov r2, r1
|
|
|
|
b sys_truncate64
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_oabi_truncate64)
|
2006-01-14 17:35:31 +01:00
|
|
|
|
|
|
|
sys_oabi_ftruncate64:
|
|
|
|
mov r3, r2
|
|
|
|
mov r2, r1
|
|
|
|
b sys_ftruncate64
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_oabi_ftruncate64)
|
2006-01-14 17:35:31 +01:00
|
|
|
|
|
|
|
sys_oabi_readahead:
|
|
|
|
str r3, [sp]
|
|
|
|
mov r3, r2
|
|
|
|
mov r2, r1
|
|
|
|
b sys_readahead
|
2008-08-28 12:22:32 +02:00
|
|
|
ENDPROC(sys_oabi_readahead)
|
2006-01-14 17:35:31 +01:00
|
|
|
|
2006-01-14 17:36:12 +01:00
|
|
|
/*
|
|
|
|
* Let's declare a second syscall table for old ABI binaries
|
|
|
|
* using the compatibility syscall entries.
|
|
|
|
*/
|
2016-10-18 20:57:01 +02:00
|
|
|
syscall_table_start sys_oabi_call_table
|
|
|
|
#define COMPAT(nr, native, compat) syscall nr, compat
|
|
|
|
#include <calls-oabi.S>
|
|
|
|
syscall_table_end sys_oabi_call_table
|
2006-01-14 17:36:12 +01:00
|
|
|
|
2006-01-14 17:35:31 +01:00
|
|
|
#endif
|
|
|
|
|