Skip to contentSkip to conversation

riscv: patch_text: Fixup last cpu should be master

PATCH Superseded [email protected] 4 participants 8 messages Apr 6, 2022 – Apr 24, 2022

Subsystems (3)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 riscv: patch_text: Fixup last cpu should be master

From: Guo Ren <[email protected]>

These patch_text implementations are using stop_machine_cpuslocked infrastructure with atomic cpu_count. The original idea: When the master CPU patch_text, the others should wait for it. But current implementation is using the first CPU as master, which couldn't guarantee the remaining CPUs are waiting. This patch changes the last CPU as the master to solve the potential risk.

---
 arch/riscv/kernel/patch.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/riscv/kernel/patch.c +1 -1
index 0b552873a577..765004b60513 100644
@@ -104,7 +104,7 @@ static int patch_text_cb(void *data)
 	struct patch_insn *patch = data;
 	int ret = 0;
 
-	if (atomic_inc_return(&patch->cpu_count) == 1) {
+	if (atomic_inc_return(&patch->cpu_count) == num_online_cpus()) {
 		ret =
 		    patch_text_nosync(patch->addr, &patch->insn,
 					    GET_INSN_LENGTH(patch->insn));
Signature · 7 lines
2.25.1


_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv
maintainer
#1 lore raw
Quoted 9 lines · On Wed, Apr 06, 2022 at 10:16:49PM +0800, [email protected] wrote:

From: Guo Ren <[email protected]>

These patch_text implementations are using stop_machine_cpuslocked infrastructure with atomic cpu_count. The original idea: When the master CPU patch_text, the others should wait for it. But current implementation is using the first CPU as master, which couldn't guarantee the remaining CPUs are waiting. This patch changes the last CPU as the master to solve the potential risk.

Quoted 8 lines
---
 arch/riscv/kernel/patch.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

What commit id does this change fix?

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

#2 lore raw
Quoted 19 lines · Greg KH
Quoted 9 lines · On Wed, Apr 06, 2022 at 10:16:49PM +0800, [email protected] wrote:

From: Guo Ren <[email protected]>

These patch_text implementations are using stop_machine_cpuslocked infrastructure with atomic cpu_count. The original idea: When the master CPU patch_text, the others should wait for it. But current implementation is using the first CPU as master, which couldn't guarantee the remaining CPUs are waiting. This patch changes the last CPU as the master to solve the potential risk.

Quoted 8 lines
---
 arch/riscv/kernel/patch.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 1 line

What commit id does this change fix?

I think it's been there since the beginning of our text patching, so

Fixes 043cb41a85de ("riscv: introduce interfaces to patch kernel code")

seems like the best bet, but I'll go take another look before merging it. That's confusing here, as I acked it, but that was for an earlier version that touched more than one arch so it was more ambiguous as to which tree it was going through (IIRC I said one of those "LMK if you want it through my tree, but here's an Ack in case someone else wants to take it" sort of things, as I usually do when it's ambiguous).

Without a changelog, cover letter, or the other patches in the set it's kind of hard to tell, though ;)

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

author
#3 lore raw
Quoted 21 lines · Palmer Dabbelt
Quoted 19 lines · Greg KH
Quoted 17 lines · On Wed, Apr 06, 2022 at 10:16:49PM +0800, [email protected] wrote:

From: Guo Ren <[email protected]>

These patch_text implementations are using stop_machine_cpuslocked infrastructure with atomic cpu_count. The original idea: When the master CPU patch_text, the others should wait for it. But current implementation is using the first CPU as master, which couldn't guarantee the remaining CPUs are waiting. This patch changes the last CPU as the master to solve the potential risk.

---
 arch/riscv/kernel/patch.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 5 lines
Quoted 1 line

What commit id does this change fix?

I think it's been there since the beginning of our text patching, so

Fixes 043cb41a85de ("riscv: introduce interfaces to patch kernel code")

Yes, it the riscv origin.

Quoted 7 lines

seems like the best bet, but I'll go take another look before merging it. That's confusing here, as I acked it, but that was for an earlier version that touched more than one arch so it was more ambiguous as to which tree it was going through (IIRC I said one of those "LMK if you want it through my tree, but here's an Ack in case someone else wants to take it" sort of things, as I usually do when it's ambiguous).

Thx for the clarification, I would remove the acked in the next version.

Quoted 3 lines

Without a changelog, cover letter, or the other patches in the set it's kind of hard to tell, though ;)

Okay, I should add a changelog for the patch with cover letter.

Signature · 9 lines
Best Regards
 Guo Ren

ML: https://lore.kernel.org/linux-csky/

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv
author
#4 lore raw
In reply to Greg KH
Quoted 20 lines · Greg KH
Quoted 9 lines · On Wed, Apr 06, 2022 at 10:16:49PM +0800, [email protected] wrote:

From: Guo Ren <[email protected]>

These patch_text implementations are using stop_machine_cpuslocked infrastructure with atomic cpu_count. The original idea: When the master CPU patch_text, the others should wait for it. But current implementation is using the first CPU as master, which couldn't guarantee the remaining CPUs are waiting. This patch changes the last CPU as the master to solve the potential risk.

Quoted 8 lines
---
 arch/riscv/kernel/patch.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 1 line

What commit id does this change fix?

Thx for pointing this out, I would follow the rule to add Cc:
<[email protected]>.

Quoted 1 line
Signature · 4 lines
Best Regards
 Guo Ren

ML: https://lore.kernel.org/linux-csky/
maintainer
#5 lore raw
In reply to [email protected]
Quoted 23 lines · On Wed, 06 Apr 2022 07:16:49 PDT (-0700), [email protected] wrote:

From: Guo Ren <[email protected]>

These patch_text implementations are using stop_machine_cpuslocked infrastructure with atomic cpu_count. The original idea: When the master CPU patch_text, the others should wait for it. But current implementation is using the first CPU as master, which couldn't guarantee the remaining CPUs are waiting. This patch changes the last CPU as the master to solve the potential risk.

---
 arch/riscv/kernel/patch.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/riscv/kernel/patch.c +0 -0
index 0b552873a577..765004b60513 100644

arch/riscv/kernel/patch.c static int patch_text_cb(void *data) · line 110

Quoted 8 lines
arch/riscv/kernel/patch.c +1 -1
@@ -104,7 +104,7 @@ static int patch_text_cb(void *data)
 	struct patch_insn *patch = data;
 	int ret = 0;
 
-	if (atomic_inc_return(&patch->cpu_count) == 1) {
+	if (atomic_inc_return(&patch->cpu_count) == num_online_cpus()) {
 		ret =
 		    patch_text_nosync(patch->addr, &patch->insn,
 					    GET_INSN_LENGTH(patch->insn));

Thanks, this is on fixes.

maintainer
#6 lore raw
Quoted 27 lines · Palmer Dabbelt
Quoted 26 lines · On Wed, 06 Apr 2022 07:16:49 PDT (-0700), [email protected] wrote:

From: Guo Ren <[email protected]>

These patch_text implementations are using stop_machine_cpuslocked infrastructure with atomic cpu_count. The original idea: When the master CPU patch_text, the others should wait for it. But current implementation is using the first CPU as master, which couldn't guarantee the remaining CPUs are waiting. This patch changes the last CPU as the master to solve the potential risk.

---
 arch/riscv/kernel/patch.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/riscv/kernel/patch.c +0 -0
index 0b552873a577..765004b60513 100644
@@ -104,7 +104,7 @@ static int patch_text_cb(void *data)
 	struct patch_insn *patch = data;
 	int ret = 0;
 
Quoted 7 lines

arch/riscv/kernel/patch.c static int patch_text_cb(void *data) · line 110

Quoted 5 lines
arch/riscv/kernel/patch.c +1 -1
-	if (atomic_inc_return(&patch->cpu_count) == 1) {
+	if (atomic_inc_return(&patch->cpu_count) == num_online_cpus()) {
 		ret =
 		    patch_text_nosync(patch->addr, &patch->insn,
 					    GET_INSN_LENGTH(patch->insn));

Thanks, this is on fixes.

Sorry, I forgot to add the Fixes and stable tags. I just fixed that up, but I'm going to hold off on this one until next week's PR to make sure it has time to go through linux-next.

author
#7 lore raw
Quoted 37 lines · Palmer Dabbelt
Quoted 27 lines · Palmer Dabbelt
Quoted 26 lines · On Wed, 06 Apr 2022 07:16:49 PDT (-0700), [email protected] wrote:

From: Guo Ren <[email protected]>

These patch_text implementations are using stop_machine_cpuslocked infrastructure with atomic cpu_count. The original idea: When the master CPU patch_text, the others should wait for it. But current implementation is using the first CPU as master, which couldn't guarantee the remaining CPUs are waiting. This patch changes the last CPU as the master to solve the potential risk.

---
 arch/riscv/kernel/patch.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/riscv/kernel/patch.c +0 -0
index 0b552873a577..765004b60513 100644
@@ -104,7 +104,7 @@ static int patch_text_cb(void *data)
     struct patch_insn *patch = data;
     int ret = 0;
 
Quoted 7 lines

arch/riscv/kernel/patch.c static int patch_text_cb(void *data) · line 110

Quoted 5 lines
arch/riscv/kernel/patch.c +1 -1
-    if (atomic_inc_return(&patch->cpu_count) == 1) {
+    if (atomic_inc_return(&patch->cpu_count) == num_online_cpus()) {
             ret =
                 patch_text_nosync(patch->addr, &patch->insn,
                                         GET_INSN_LENGTH(patch->insn));

Thanks, this is on fixes.

Quoted 3 lines

Sorry, I forgot to add the Fixes and stable tags. I just fixed that up, but I'm going to hold off on this one until next week's PR to make sure it has time to go through linux-next.

https://lore.kernel.org/linux-riscv/[email protected]/

Signature · 4 lines
Best Regards
 Guo Ren

ML: https://lore.kernel.org/linux-csky/