Re: [PATCH v5 10/13] powerpc/ptrace: split out ADV_DEBUG_REGS related functions.
Le 24/03/2020 à 07:23, Michael Ellerman a écrit :
Christophe Leroy writes:
On 03/20/2020 02:12 AM, Michael Ellerman wrote:
Christophe Leroy writes:
Move ADV_DEBUG_REGS functions out of ptrace.c, into
ptrace-adv.c and ptrace-noadv.c
Signed-off-by: Christophe Leroy
---
v4: Leave hw_breakpoint.h for ptrace.c
---
arch/powerpc/kernel/ptrace/Makefile | 4 +
arch/powerpc/kernel/ptrace/ptrace-adv.c | 468
arch/powerpc/kernel/ptrace/ptrace-decl.h | 5 +
arch/powerpc/kernel/ptrace/ptrace-noadv.c | 236
arch/powerpc/kernel/ptrace/ptrace.c | 650 --
5 files changed, 713 insertions(+), 650 deletions(-)
create mode 100644 arch/powerpc/kernel/ptrace/ptrace-adv.c
create mode 100644 arch/powerpc/kernel/ptrace/ptrace-noadv.c
This is somehow breaking the ptrace-hwbreak selftest on Power8:
test: ptrace-hwbreak
tags: git_version:v5.6-rc6-892-g7a285a6067d6
PTRACE_SET_DEBUGREG, WO, len: 1: Ok
PTRACE_SET_DEBUGREG, WO, len: 2: Ok
PTRACE_SET_DEBUGREG, WO, len: 4: Ok
PTRACE_SET_DEBUGREG, WO, len: 8: Ok
PTRACE_SET_DEBUGREG, RO, len: 1: Ok
PTRACE_SET_DEBUGREG, RO, len: 2: Ok
PTRACE_SET_DEBUGREG, RO, len: 4: Ok
PTRACE_SET_DEBUGREG, RO, len: 8: Ok
PTRACE_SET_DEBUGREG, RW, len: 1: Ok
PTRACE_SET_DEBUGREG, RW, len: 2: Ok
PTRACE_SET_DEBUGREG, RW, len: 4: Ok
PTRACE_SET_DEBUGREG, RW, len: 8: Ok
PPC_PTRACE_SETHWDEBUG, MODE_EXACT, WO, len: 1: Ok
PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RO, len: 1: Ok
PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RW, len: 1: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, WO, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RO, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RW, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, WO, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, RO, len: 6: Fail
failure: ptrace-hwbreak
I haven't had time to work out why yet.
A (big) part of commit c3f68b0478e7 ("powerpc/watchpoint: Fix ptrace
code that muck around with address/len") was lost during rebase.
I'll send a fix, up to you to squash it in or commit it as is.
Thanks.
One person asked me if I sent the fix already. So yes, it is there:
https://patchwork.ozlabs.org/patch/1259348/
Christophe
Re: [PATCH v5 10/13] powerpc/ptrace: split out ADV_DEBUG_REGS related functions.
Christophe Leroy writes:
> On 03/20/2020 02:12 AM, Michael Ellerman wrote:
>> Christophe Leroy writes:
>>> Move ADV_DEBUG_REGS functions out of ptrace.c, into
>>> ptrace-adv.c and ptrace-noadv.c
>>>
>>> Signed-off-by: Christophe Leroy
>>> ---
>>> v4: Leave hw_breakpoint.h for ptrace.c
>>> ---
>>> arch/powerpc/kernel/ptrace/Makefile | 4 +
>>> arch/powerpc/kernel/ptrace/ptrace-adv.c | 468
>>> arch/powerpc/kernel/ptrace/ptrace-decl.h | 5 +
>>> arch/powerpc/kernel/ptrace/ptrace-noadv.c | 236
>>> arch/powerpc/kernel/ptrace/ptrace.c | 650 --
>>> 5 files changed, 713 insertions(+), 650 deletions(-)
>>> create mode 100644 arch/powerpc/kernel/ptrace/ptrace-adv.c
>>> create mode 100644 arch/powerpc/kernel/ptrace/ptrace-noadv.c
>>
>> This is somehow breaking the ptrace-hwbreak selftest on Power8:
>>
>>test: ptrace-hwbreak
>>tags: git_version:v5.6-rc6-892-g7a285a6067d6
>>PTRACE_SET_DEBUGREG, WO, len: 1: Ok
>>PTRACE_SET_DEBUGREG, WO, len: 2: Ok
>>PTRACE_SET_DEBUGREG, WO, len: 4: Ok
>>PTRACE_SET_DEBUGREG, WO, len: 8: Ok
>>PTRACE_SET_DEBUGREG, RO, len: 1: Ok
>>PTRACE_SET_DEBUGREG, RO, len: 2: Ok
>>PTRACE_SET_DEBUGREG, RO, len: 4: Ok
>>PTRACE_SET_DEBUGREG, RO, len: 8: Ok
>>PTRACE_SET_DEBUGREG, RW, len: 1: Ok
>>PTRACE_SET_DEBUGREG, RW, len: 2: Ok
>>PTRACE_SET_DEBUGREG, RW, len: 4: Ok
>>PTRACE_SET_DEBUGREG, RW, len: 8: Ok
>>PPC_PTRACE_SETHWDEBUG, MODE_EXACT, WO, len: 1: Ok
>>PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RO, len: 1: Ok
>>PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RW, len: 1: Ok
>>PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, WO, len: 6: Ok
>>PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RO, len: 6: Ok
>>PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RW, len: 6: Ok
>>PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, WO, len: 6: Ok
>>PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, RO, len: 6: Fail
>>failure: ptrace-hwbreak
>>
>> I haven't had time to work out why yet.
>>
>
> A (big) part of commit c3f68b0478e7 ("powerpc/watchpoint: Fix ptrace
> code that muck around with address/len") was lost during rebase.
>
> I'll send a fix, up to you to squash it in or commit it as is.
Thanks.
cheers
Re: [PATCH v5 10/13] powerpc/ptrace: split out ADV_DEBUG_REGS related functions.
On 03/20/2020 02:12 AM, Michael Ellerman wrote:
Christophe Leroy writes:
Move ADV_DEBUG_REGS functions out of ptrace.c, into
ptrace-adv.c and ptrace-noadv.c
Signed-off-by: Christophe Leroy
---
v4: Leave hw_breakpoint.h for ptrace.c
---
arch/powerpc/kernel/ptrace/Makefile | 4 +
arch/powerpc/kernel/ptrace/ptrace-adv.c | 468
arch/powerpc/kernel/ptrace/ptrace-decl.h | 5 +
arch/powerpc/kernel/ptrace/ptrace-noadv.c | 236
arch/powerpc/kernel/ptrace/ptrace.c | 650 --
5 files changed, 713 insertions(+), 650 deletions(-)
create mode 100644 arch/powerpc/kernel/ptrace/ptrace-adv.c
create mode 100644 arch/powerpc/kernel/ptrace/ptrace-noadv.c
This is somehow breaking the ptrace-hwbreak selftest on Power8:
test: ptrace-hwbreak
tags: git_version:v5.6-rc6-892-g7a285a6067d6
PTRACE_SET_DEBUGREG, WO, len: 1: Ok
PTRACE_SET_DEBUGREG, WO, len: 2: Ok
PTRACE_SET_DEBUGREG, WO, len: 4: Ok
PTRACE_SET_DEBUGREG, WO, len: 8: Ok
PTRACE_SET_DEBUGREG, RO, len: 1: Ok
PTRACE_SET_DEBUGREG, RO, len: 2: Ok
PTRACE_SET_DEBUGREG, RO, len: 4: Ok
PTRACE_SET_DEBUGREG, RO, len: 8: Ok
PTRACE_SET_DEBUGREG, RW, len: 1: Ok
PTRACE_SET_DEBUGREG, RW, len: 2: Ok
PTRACE_SET_DEBUGREG, RW, len: 4: Ok
PTRACE_SET_DEBUGREG, RW, len: 8: Ok
PPC_PTRACE_SETHWDEBUG, MODE_EXACT, WO, len: 1: Ok
PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RO, len: 1: Ok
PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RW, len: 1: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, WO, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RO, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RW, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, WO, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, RO, len: 6: Fail
failure: ptrace-hwbreak
I haven't had time to work out why yet.
A (big) part of commit c3f68b0478e7 ("powerpc/watchpoint: Fix ptrace
code that muck around with address/len") was lost during rebase.
I'll send a fix, up to you to squash it in or commit it as is.
Christophe
Re: [PATCH v5 10/13] powerpc/ptrace: split out ADV_DEBUG_REGS related functions.
Christophe Leroy writes:
> Move ADV_DEBUG_REGS functions out of ptrace.c, into
> ptrace-adv.c and ptrace-noadv.c
>
> Signed-off-by: Christophe Leroy
> ---
> v4: Leave hw_breakpoint.h for ptrace.c
> ---
> arch/powerpc/kernel/ptrace/Makefile | 4 +
> arch/powerpc/kernel/ptrace/ptrace-adv.c | 468
> arch/powerpc/kernel/ptrace/ptrace-decl.h | 5 +
> arch/powerpc/kernel/ptrace/ptrace-noadv.c | 236
> arch/powerpc/kernel/ptrace/ptrace.c | 650 --
> 5 files changed, 713 insertions(+), 650 deletions(-)
> create mode 100644 arch/powerpc/kernel/ptrace/ptrace-adv.c
> create mode 100644 arch/powerpc/kernel/ptrace/ptrace-noadv.c
This is somehow breaking the ptrace-hwbreak selftest on Power8:
test: ptrace-hwbreak
tags: git_version:v5.6-rc6-892-g7a285a6067d6
PTRACE_SET_DEBUGREG, WO, len: 1: Ok
PTRACE_SET_DEBUGREG, WO, len: 2: Ok
PTRACE_SET_DEBUGREG, WO, len: 4: Ok
PTRACE_SET_DEBUGREG, WO, len: 8: Ok
PTRACE_SET_DEBUGREG, RO, len: 1: Ok
PTRACE_SET_DEBUGREG, RO, len: 2: Ok
PTRACE_SET_DEBUGREG, RO, len: 4: Ok
PTRACE_SET_DEBUGREG, RO, len: 8: Ok
PTRACE_SET_DEBUGREG, RW, len: 1: Ok
PTRACE_SET_DEBUGREG, RW, len: 2: Ok
PTRACE_SET_DEBUGREG, RW, len: 4: Ok
PTRACE_SET_DEBUGREG, RW, len: 8: Ok
PPC_PTRACE_SETHWDEBUG, MODE_EXACT, WO, len: 1: Ok
PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RO, len: 1: Ok
PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RW, len: 1: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, WO, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RO, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RW, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, WO, len: 6: Ok
PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, RO, len: 6: Fail
failure: ptrace-hwbreak
I haven't had time to work out why yet.
cheers
> diff --git a/arch/powerpc/kernel/ptrace/Makefile
> b/arch/powerpc/kernel/ptrace/Makefile
> index 7addc5994bb9..e9d97c2d063e 100644
> --- a/arch/powerpc/kernel/ptrace/Makefile
> +++ b/arch/powerpc/kernel/ptrace/Makefile
> @@ -14,3 +14,7 @@ endif
> obj-$(CONFIG_ALTIVEC)+= ptrace-altivec.o
> obj-$(CONFIG_SPE)+= ptrace-spe.o
> obj-$(CONFIG_PPC_TRANSACTIONAL_MEM) += ptrace-tm.o
> +obj-$(CONFIG_PPC_ADV_DEBUG_REGS) += ptrace-adv.o
> +ifneq ($(CONFIG_PPC_ADV_DEBUG_REGS),y)
> +obj-y+= ptrace-noadv.o
> +endif
> diff --git a/arch/powerpc/kernel/ptrace/ptrace-adv.c
> b/arch/powerpc/kernel/ptrace/ptrace-adv.c
> new file mode 100644
> index ..eebcd41edc3d
> --- /dev/null
> +++ b/arch/powerpc/kernel/ptrace/ptrace-adv.c
> @@ -0,0 +1,468 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +
> +#include
> +#include
> +
> +#include "ptrace-decl.h"
> +
> +void user_enable_single_step(struct task_struct *task)
> +{
> + struct pt_regs *regs = task->thread.regs;
> +
> + if (regs != NULL) {
> + task->thread.debug.dbcr0 &= ~DBCR0_BT;
> + task->thread.debug.dbcr0 |= DBCR0_IDM | DBCR0_IC;
> + regs->msr |= MSR_DE;
> + }
> + set_tsk_thread_flag(task, TIF_SINGLESTEP);
> +}
> +
> +void user_enable_block_step(struct task_struct *task)
> +{
> + struct pt_regs *regs = task->thread.regs;
> +
> + if (regs != NULL) {
> + task->thread.debug.dbcr0 &= ~DBCR0_IC;
> + task->thread.debug.dbcr0 = DBCR0_IDM | DBCR0_BT;
> + regs->msr |= MSR_DE;
> + }
> + set_tsk_thread_flag(task, TIF_SINGLESTEP);
> +}
> +
> +void user_disable_single_step(struct task_struct *task)
> +{
> + struct pt_regs *regs = task->thread.regs;
> +
> + if (regs != NULL) {
> + /*
> + * The logic to disable single stepping should be as
> + * simple as turning off the Instruction Complete flag.
> + * And, after doing so, if all debug flags are off, turn
> + * off DBCR0(IDM) and MSR(DE) Torez
> + */
> + task->thread.debug.dbcr0 &= ~(DBCR0_IC | DBCR0_BT);
> + /*
> + * Test to see if any of the DBCR_ACTIVE_EVENTS bits are set.
> + */
> + if (!DBCR_ACTIVE_EVENTS(task->thread.debug.dbcr0,
> + task->thread.debug.dbcr1)) {
> + /*
> + * All debug events were off.
> + */
> + task->thread.debug.dbcr0 &= ~DBCR0_IDM;
> + regs->msr &= ~MSR_DE;
> + }
> + }
> + clear_tsk_thread_flag(task, TIF_SINGLESTEP);
> +}
> +
> +int ptrace_set_debugreg(struct task_struct *task, unsigned long addr,
> unsigned long data)
> +{
> +#ifdef CONFIG_HAVE_HW_BREAKPOINT
> + int ret;
> + struct thread_struct *thread = >thread;
> + struct perf_event *bp;
> + struct perf_event_attr attr;
> +#endif /* CONFIG_HAVE_HW_BREAKPOINT */
> +
> + /* For ppc64 we support one DABR and no IABR's at the moment
[PATCH v5 10/13] powerpc/ptrace: split out ADV_DEBUG_REGS related functions.
Move ADV_DEBUG_REGS functions out of ptrace.c, into
ptrace-adv.c and ptrace-noadv.c
Signed-off-by: Christophe Leroy
---
v4: Leave hw_breakpoint.h for ptrace.c
---
arch/powerpc/kernel/ptrace/Makefile | 4 +
arch/powerpc/kernel/ptrace/ptrace-adv.c | 468
arch/powerpc/kernel/ptrace/ptrace-decl.h | 5 +
arch/powerpc/kernel/ptrace/ptrace-noadv.c | 236
arch/powerpc/kernel/ptrace/ptrace.c | 650 --
5 files changed, 713 insertions(+), 650 deletions(-)
create mode 100644 arch/powerpc/kernel/ptrace/ptrace-adv.c
create mode 100644 arch/powerpc/kernel/ptrace/ptrace-noadv.c
diff --git a/arch/powerpc/kernel/ptrace/Makefile
b/arch/powerpc/kernel/ptrace/Makefile
index 7addc5994bb9..e9d97c2d063e 100644
--- a/arch/powerpc/kernel/ptrace/Makefile
+++ b/arch/powerpc/kernel/ptrace/Makefile
@@ -14,3 +14,7 @@ endif
obj-$(CONFIG_ALTIVEC) += ptrace-altivec.o
obj-$(CONFIG_SPE) += ptrace-spe.o
obj-$(CONFIG_PPC_TRANSACTIONAL_MEM)+= ptrace-tm.o
+obj-$(CONFIG_PPC_ADV_DEBUG_REGS) += ptrace-adv.o
+ifneq ($(CONFIG_PPC_ADV_DEBUG_REGS),y)
+obj-y += ptrace-noadv.o
+endif
diff --git a/arch/powerpc/kernel/ptrace/ptrace-adv.c
b/arch/powerpc/kernel/ptrace/ptrace-adv.c
new file mode 100644
index ..eebcd41edc3d
--- /dev/null
+++ b/arch/powerpc/kernel/ptrace/ptrace-adv.c
@@ -0,0 +1,468 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#include
+#include
+
+#include "ptrace-decl.h"
+
+void user_enable_single_step(struct task_struct *task)
+{
+ struct pt_regs *regs = task->thread.regs;
+
+ if (regs != NULL) {
+ task->thread.debug.dbcr0 &= ~DBCR0_BT;
+ task->thread.debug.dbcr0 |= DBCR0_IDM | DBCR0_IC;
+ regs->msr |= MSR_DE;
+ }
+ set_tsk_thread_flag(task, TIF_SINGLESTEP);
+}
+
+void user_enable_block_step(struct task_struct *task)
+{
+ struct pt_regs *regs = task->thread.regs;
+
+ if (regs != NULL) {
+ task->thread.debug.dbcr0 &= ~DBCR0_IC;
+ task->thread.debug.dbcr0 = DBCR0_IDM | DBCR0_BT;
+ regs->msr |= MSR_DE;
+ }
+ set_tsk_thread_flag(task, TIF_SINGLESTEP);
+}
+
+void user_disable_single_step(struct task_struct *task)
+{
+ struct pt_regs *regs = task->thread.regs;
+
+ if (regs != NULL) {
+ /*
+* The logic to disable single stepping should be as
+* simple as turning off the Instruction Complete flag.
+* And, after doing so, if all debug flags are off, turn
+* off DBCR0(IDM) and MSR(DE) Torez
+*/
+ task->thread.debug.dbcr0 &= ~(DBCR0_IC | DBCR0_BT);
+ /*
+* Test to see if any of the DBCR_ACTIVE_EVENTS bits are set.
+*/
+ if (!DBCR_ACTIVE_EVENTS(task->thread.debug.dbcr0,
+ task->thread.debug.dbcr1)) {
+ /*
+* All debug events were off.
+*/
+ task->thread.debug.dbcr0 &= ~DBCR0_IDM;
+ regs->msr &= ~MSR_DE;
+ }
+ }
+ clear_tsk_thread_flag(task, TIF_SINGLESTEP);
+}
+
+int ptrace_set_debugreg(struct task_struct *task, unsigned long addr, unsigned
long data)
+{
+#ifdef CONFIG_HAVE_HW_BREAKPOINT
+ int ret;
+ struct thread_struct *thread = >thread;
+ struct perf_event *bp;
+ struct perf_event_attr attr;
+#endif /* CONFIG_HAVE_HW_BREAKPOINT */
+
+ /* For ppc64 we support one DABR and no IABR's at the moment (ppc64).
+* For embedded processors we support one DAC and no IAC's at the
+* moment.
+*/
+ if (addr > 0)
+ return -EINVAL;
+
+ /* The bottom 3 bits in dabr are flags */
+ if ((data & ~0x7UL) >= TASK_SIZE)
+ return -EIO;
+
+ /* As described above, it was assumed 3 bits were passed with the data
+* address, but we will assume only the mode bits will be passed
+* as to not cause alignment restrictions for DAC-based processors.
+*/
+
+ /* DAC's hold the whole address without any mode flags */
+ task->thread.debug.dac1 = data & ~0x3UL;
+
+ if (task->thread.debug.dac1 == 0) {
+ dbcr_dac(task) &= ~(DBCR_DAC1R | DBCR_DAC1W);
+ if (!DBCR_ACTIVE_EVENTS(task->thread.debug.dbcr0,
+ task->thread.debug.dbcr1)) {
+ task->thread.regs->msr &= ~MSR_DE;
+ task->thread.debug.dbcr0 &= ~DBCR0_IDM;
+ }
+ return 0;
+ }
+
+ /* Read or Write bits must be set */
+
+ if (!(data & 0x3UL))
+ return -EINVAL;
+
+ /* Set the Internal Debugging flag (IDM bit 1) for the DBCR0 register */
+ task->thread.debug.dbcr0 |=
