From: Andr? Hentschel <[email protected]>
Since commit 6a1c53124aa1 the user writeable TLS register was zeroed to
prevent it from being used as a covert channel between two tasks.
There are more and more applications coming to Windows RT,
Wine could support them, but mostly they expect to have
the thread environment block (TEB) in TPIDRURW.
This patch preserves that register per thread instead of clearing it.
Unlike the TPIDRURO, which is already switched, the TPIDRURW
can be updated from userspace so needs careful treatment in the case that we
modify TPIDRURW and call fork(). To avoid this we must always read
TPIDRURW in copy_thread.
Signed-off-by: Andr? Hentschel <[email protected]>
Signed-off-by: Will Deacon <[email protected]>
Signed-off-by: Jonathan Austin <[email protected]>
---
This patch is against Linux 3.10-rc6 (7d132055814ef17a6c7b69f342244c410a5e000f)
v2: rework and fixup of v1, based on a suggested patch by Will Deacon
v3: total rework and fixup of v2
v4: removed condition on assembler instruction,
adapted my code to kernel-style, both based on comments by Will Deacon
v5: rebased v4 on 3.10-rc2 and adding this version history
v6: moved loading the TLS registers to the macros
(fixing the "LDRD is not supported on all the CPUs we have" problem)
Why so much Signed-off-bys? Some History:
The first patch had performance issues pointed out by Russel King,
so Will Deacon jumped in to help me with that. The second one again
had performance issues and the missing copy_thread part was uncovered.
After some iterations by me, Jonathan Austin proposed a patch and
Russel King sent his idea of the assembler part. All this was finally
merged and refined into this patch.
Thanks to everyone!
diff --git a/arch/arm/include/asm/thread_info.h b/arch/arm/include/asm/thread_info.h
index 1995d1a..214d415 100644
--- a/arch/arm/include/asm/thread_info.h
+++ b/arch/arm/include/asm/thread_info.h
@@ -58,7 +58,7 @@ struct thread_info {
struct cpu_context_save cpu_context; /* cpu context */
__u32 syscall; /* syscall number */
__u8 used_cp[16]; /* thread used copro */
- unsigned long tp_value;
+ unsigned long tp_value[2]; /* TLS registers */
#ifdef CONFIG_CRUNCH
struct crunch_state crunchstate;
#endif
diff --git a/arch/arm/include/asm/tls.h b/arch/arm/include/asm/tls.h
index 73409e6..3742722 100644
--- a/arch/arm/include/asm/tls.h
+++ b/arch/arm/include/asm/tls.h
@@ -2,27 +2,33 @@
#define __ASMARM_TLS_H
#ifdef __ASSEMBLY__
- .macro set_tls_none, tp, tmp1, tmp2
+#include <asm/asm-offsets.h>
+ .macro switch_tls_none, prev, next, tp, tpuser, tmp1, tmp2
.endm
- .macro set_tls_v6k, tp, tmp1, tmp2
+ .macro switch_tls_v6k, prev, next, tp, tpuser, tmp1, tmp2
+ ldrd \tp, \tpuser, [\next, #TI_TP_VALUE] @ get the next TLS and user r/w register
+ mrc p15, 0, \tmp2, c13, c0, 2 @ get the user r/w register
mcr p15, 0, \tp, c13, c0, 3 @ set TLS register
- mov \tmp1, #0
- mcr p15, 0, \tmp1, c13, c0, 2 @ clear user r/w TLS register
+ mcr p15, 0, \tpuser, c13, c0, 2 @ and the user r/w register
+ str \tmp2, [\prev, #TI_TP_VALUE + 4] @ save it
.endm
- .macro set_tls_v6, tp, tmp1, tmp2
+ .macro switch_tls_v6, prev, next, tp, tpuser, tmp1, tmp2
ldr \tmp1, =elf_hwcap
ldr \tmp1, [\tmp1, #0]
mov \tmp2, #0xffff0fff
+ ldr \tp, [\next, #TI_TP_VALUE] @ get the next TLS register
tst \tmp1, #HWCAP_TLS @ hardware TLS available?
- mcrne p15, 0, \tp, c13, c0, 3 @ yes, set TLS register
- movne \tmp1, #0
- mcrne p15, 0, \tmp1, c13, c0, 2 @ clear user r/w TLS register
streq \tp, [\tmp2, #-15] @ set TLS value at 0xffff0ff0
+ mrcne p15, 0, \tmp2, c13, c0, 2 @ get the previous user r/w register
+ ldrne \tpuser, [\next, #TI_TP_VALUE + 4] @ get the next user r/w register
+ mcrne p15, 0, \tp, c13, c0, 3 @ yes, set TLS register
+ mcrne p15, 0, \tpuser, c13, c0, 2 @ set user r/w register
+ strne \tmp2, [\prev, #TI_TP_VALUE + 4] @ save it
.endm
- .macro set_tls_software, tp, tmp1, tmp2
+ .macro switch_tls_software, prev, next, tp, tpuser, tmp1, tmp2
mov \tmp1, #0xffff0fff
str \tp, [\tmp1, #-15] @ set TLS value at 0xffff0ff0
.endm
@@ -31,19 +37,30 @@
#ifdef CONFIG_TLS_REG_EMUL
#define tls_emu 1
#define has_tls_reg 1
-#define set_tls set_tls_none
+#define switch_tls switch_tls_none
#elif defined(CONFIG_CPU_V6)
#define tls_emu 0
#define has_tls_reg (elf_hwcap & HWCAP_TLS)
-#define set_tls set_tls_v6
+#define switch_tls switch_tls_v6
#elif defined(CONFIG_CPU_32v6K)
#define tls_emu 0
#define has_tls_reg 1
-#define set_tls set_tls_v6k
+#define switch_tls switch_tls_v6k
#else
#define tls_emu 0
#define has_tls_reg 0
-#define set_tls set_tls_software
+#define switch_tls switch_tls_software
#endif
+#ifndef __ASSEMBLY__
+static inline unsigned long get_tpuser(void)
+{
+ unsigned long reg = 0;
+
+ if (has_tls_reg && !tls_emu)
+ __asm__("mrc p15, 0, %0, c13, c0, 2" : "=r" (reg));
+
+ return reg;
+}
+#endif
#endif /* __ASMARM_TLS_H */
diff --git a/arch/arm/kernel/entry-armv.S b/arch/arm/kernel/entry-armv.S
index 582b405..1484b59 100644
--- a/arch/arm/kernel/entry-armv.S
+++ b/arch/arm/kernel/entry-armv.S
@@ -685,7 +685,6 @@ ENTRY(__switch_to)
UNWIND(.fnstart )
UNWIND(.cantunwind )
add ip, r1, #TI_CPU_SAVE
- ldr r3, [r2, #TI_TP_VALUE]
ARM( stmia ip!, {r4 - sl, fp, sp, lr} ) @ Store most regs on stack
THUMB( stmia ip!, {r4 - sl, fp} ) @ Store most regs on stack
THUMB( str sp, [ip], #4 )
@@ -693,7 +692,7 @@ ENTRY(__switch_to)
#ifdef CONFIG_CPU_USE_DOMAINS
ldr r6, [r2, #TI_CPU_DOMAIN]
#endif
- set_tls r3, r4, r5
+ switch_tls r1, r2, r4, r5, r3, r7
#if defined(CONFIG_CC_STACKPROTECTOR) && !defined(CONFIG_SMP)
ldr r7, [r2, #TI_TASK]
ldr r8, =__stack_chk_guard
diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c
index 282de48..a24110f 100644
--- a/arch/arm/kernel/process.c
+++ b/arch/arm/kernel/process.c
@@ -39,6 +39,7 @@
#include <asm/thread_notify.h>
#include <asm/stacktrace.h>
#include <asm/mach/time.h>
+#include <asm/tls.h>
#ifdef CONFIG_CC_STACKPROTECTOR
#include <linux/stackprotector.h>
@@ -343,7 +344,8 @@ copy_thread(unsigned long clone_flags, unsigned long stack_start,
clear_ptrace_hw_breakpoint(p);
if (clone_flags & CLONE_SETTLS)
- thread->tp_value = childregs->ARM_r3;
+ thread->tp_value[0] = childregs->ARM_r3;
+ thread->tp_value[1] = get_tpuser();
thread_notify(THREAD_NOTIFY_COPY, thread);
diff --git a/arch/arm/kernel/ptrace.c b/arch/arm/kernel/ptrace.c
index 03deeff..2bc1514 100644
--- a/arch/arm/kernel/ptrace.c
+++ b/arch/arm/kernel/ptrace.c
@@ -849,7 +849,7 @@ long arch_ptrace(struct task_struct *child, long request,
#endif
case PTRACE_GET_THREAD_AREA:
- ret = put_user(task_thread_info(child)->tp_value,
+ ret = put_user(task_thread_info(child)->tp_value[0],
datap);
break;
diff --git a/arch/arm/kernel/traps.c b/arch/arm/kernel/traps.c
index 18b32e8..517bfd4 100644
--- a/arch/arm/kernel/traps.c
+++ b/arch/arm/kernel/traps.c
@@ -581,7 +581,7 @@ asmlinkage int arm_syscall(int no, struct pt_regs *regs)
return regs->ARM_r0;
case NR(set_tls):
- thread->tp_value = regs->ARM_r0;
+ thread->tp_value[0] = regs->ARM_r0;
if (tls_emu)
return 0;
if (has_tls_reg) {
@@ -699,7 +699,7 @@ static int get_tp_trap(struct pt_regs *regs, unsigned int instr)
int reg = (instr >> 12) & 15;
if (reg == 15)
return 1;
- regs->uregs[reg] = current_thread_info()->tp_value;
+ regs->uregs[reg] = current_thread_info()->tp_value[0];
regs->ARM_pc += 4;
return 0;
}
--
1.8.1.2
On Mon, Jun 17, 2013 at 11:13:48PM +0100, Andr? Hentschel wrote:
> From: Andr? Hentschel <[email protected]>
>
> Since commit 6a1c53124aa1 the user writeable TLS register was zeroed to
> prevent it from being used as a covert channel between two tasks.
>
> There are more and more applications coming to Windows RT,
> Wine could support them, but mostly they expect to have
> the thread environment block (TEB) in TPIDRURW.
>
> This patch preserves that register per thread instead of clearing it.
> Unlike the TPIDRURO, which is already switched, the TPIDRURW
> can be updated from userspace so needs careful treatment in the case that we
> modify TPIDRURW and call fork(). To avoid this we must always read
> TPIDRURW in copy_thread.
>
> Signed-off-by: Andr? Hentschel <[email protected]>
> Signed-off-by: Will Deacon <[email protected]>
> Signed-off-by: Jonathan Austin <[email protected]>
>
> ---
> This patch is against Linux 3.10-rc6 (7d132055814ef17a6c7b69f342244c410a5e000f)
>
> v2: rework and fixup of v1, based on a suggested patch by Will Deacon
> v3: total rework and fixup of v2
> v4: removed condition on assembler instruction,
> adapted my code to kernel-style, both based on comments by Will Deacon
> v5: rebased v4 on 3.10-rc2 and adding this version history
> v6: moved loading the TLS registers to the macros
> (fixing the "LDRD is not supported on all the CPUs we have" problem)
You've changed quite a lot with this version, including the way the macro
parameters are passed. Why not just replace the problematic ldrd with two
ldr instructions and be done with it? I don't think the simple build error
warrants an overhaul of the code we already had.
Cheers,
Will
On 18.06.2013 12:07, Will Deacon wrote:
> On Mon, Jun 17, 2013 at 11:13:48PM +0100, Andr? Hentschel wrote:
>> From: Andr? Hentschel <[email protected]>
>>
>> Since commit 6a1c53124aa1 the user writeable TLS register was zeroed to
>> prevent it from being used as a covert channel between two tasks.
>>
>> There are more and more applications coming to Windows RT,
>> Wine could support them, but mostly they expect to have
>> the thread environment block (TEB) in TPIDRURW.
>>
>> This patch preserves that register per thread instead of clearing it.
>> Unlike the TPIDRURO, which is already switched, the TPIDRURW
>> can be updated from userspace so needs careful treatment in the case that we
>> modify TPIDRURW and call fork(). To avoid this we must always read
>> TPIDRURW in copy_thread.
>>
>> Signed-off-by: Andr? Hentschel <[email protected]>
>> Signed-off-by: Will Deacon <[email protected]>
>> Signed-off-by: Jonathan Austin <[email protected]>
>>
>> ---
>> This patch is against Linux 3.10-rc6 (7d132055814ef17a6c7b69f342244c410a5e000f)
>>
>> v2: rework and fixup of v1, based on a suggested patch by Will Deacon
>> v3: total rework and fixup of v2
>> v4: removed condition on assembler instruction,
>> adapted my code to kernel-style, both based on comments by Will Deacon
>> v5: rebased v4 on 3.10-rc2 and adding this version history
>> v6: moved loading the TLS registers to the macros
>> (fixing the "LDRD is not supported on all the CPUs we have" problem)
>
> You've changed quite a lot with this version, including the way the macro
> parameters are passed. Why not just replace the problematic ldrd with two
> ldr instructions and be done with it? I don't think the simple build error
> warrants an overhaul of the code we already had.
Wantig that patch to be in 3.11 i thought i should do more now to push it.
I'm still not that familiar with the process, but i think Russell King would
have done this easy change himself when he would be happy with it.
Further this patch seems cleaner and much likely performing better.
I'm off till saturday, hopefully i'm enlighted by feedback till then :)
(just to be clear: i'm also happy with replacing ldrd with two ldr instructions)
On Tue, Jun 18, 2013 at 07:58:27PM +0100, Andr? Hentschel wrote:
> On 18.06.2013 12:07, Will Deacon wrote:
> > On Mon, Jun 17, 2013 at 11:13:48PM +0100, Andr? Hentschel wrote:
> >> From: Andr? Hentschel <[email protected]>
> >>
> >> Since commit 6a1c53124aa1 the user writeable TLS register was zeroed to
> >> prevent it from being used as a covert channel between two tasks.
> >>
> >> There are more and more applications coming to Windows RT,
> >> Wine could support them, but mostly they expect to have
> >> the thread environment block (TEB) in TPIDRURW.
> >>
> >> This patch preserves that register per thread instead of clearing it.
> >> Unlike the TPIDRURO, which is already switched, the TPIDRURW
> >> can be updated from userspace so needs careful treatment in the case that we
> >> modify TPIDRURW and call fork(). To avoid this we must always read
> >> TPIDRURW in copy_thread.
> >>
> >> Signed-off-by: Andr? Hentschel <[email protected]>
> >> Signed-off-by: Will Deacon <[email protected]>
> >> Signed-off-by: Jonathan Austin <[email protected]>
> >>
> >> ---
> >> This patch is against Linux 3.10-rc6 (7d132055814ef17a6c7b69f342244c410a5e000f)
> >>
> >> v2: rework and fixup of v1, based on a suggested patch by Will Deacon
> >> v3: total rework and fixup of v2
> >> v4: removed condition on assembler instruction,
> >> adapted my code to kernel-style, both based on comments by Will Deacon
> >> v5: rebased v4 on 3.10-rc2 and adding this version history
> >> v6: moved loading the TLS registers to the macros
> >> (fixing the "LDRD is not supported on all the CPUs we have" problem)
> >
> > You've changed quite a lot with this version, including the way the macro
> > parameters are passed. Why not just replace the problematic ldrd with two
> > ldr instructions and be done with it? I don't think the simple build error
> > warrants an overhaul of the code we already had.
>
> Wantig that patch to be in 3.11 i thought i should do more now to push it.
It's getting pretty late for 3.11, so generally you'd want to make the most
minimal changes possible, rather than redesign core parts of the patch.
> I'm still not that familiar with the process, but i think Russell King would
> have done this easy change himself when he would be happy with it.
That's unlikely -- you're still the author on the patch so I wouldn't expect
the gatekeeper to make any changes above merge conflicts.
> Further this patch seems cleaner and much likely performing better.
Well, it's not the approach that has been reviewed up until now and, without
numbers backing up your claims, I'd be inclined simply to make the simple
change of adding two ldr instructions.
Will