Skip to contentSkip to conversation

Perf degradation from -rt14 onwards

DISCUSSION Dinakar Guniguntala 3 participants 3 messages Dec 1, 2005

lore
Share & tools
lore
#0 lore raw

I was wondering why the following change was made from -rt14 onwards.

@@ -1634,7 +1531,7 @@ asmlinkage long sys_futex(u32 __user *ua
int val3)

 {
        struct timespec t;

- unsigned long timeout = MAX_SCHEDULE_TIMEOUT;
+ unsigned long timeout = 0;

This was introduced in patch-2.6.14-rt13-rf3 by David.

This seems to return spurious -ETIMEDOUT errors even in the non-robust code and results in userspace (glibc) retrying several mutex operations before it succeeds. I was chasing down a degradation of performance of some testcases and was able to fix those by reverting this change back.

-Dinakar

#1 lore raw
Quoted 13 lines · Dinakar Guniguntala

I was wondering why the following change was made from -rt14 onwards.

(quoted diff) +1 -1
@@ -1634,7 +1531,7 @@ asmlinkage long sys_futex(u32 __user *ua
                         int val3)
 {
       struct timespec t;
-       unsigned long timeout = MAX_SCHEDULE_TIMEOUT;
+       unsigned long timeout = 0;
 

This was introduced in patch-2.6.14-rt13-rf3 by David.

Quoted 7 lines

This seems to return spurious -ETIMEDOUT errors even in the non-robust code and results in userspace (glibc) retrying several mutex operations before it succeeds. I was chasing down a degradation of performance of some testcases and was able to fix those by reverting this change back.

Yes. The default timeout should be set to MAX_SCHEDULE_TIMEOUT, not zero.

David

Quoted 4 lines

-Dinakar

#2 lore raw
In reply to Dinakar Guniguntala
Quoted 13 lines · Dinakar Guniguntala

I was wondering why the following change was made from -rt14 onwards.

(quoted diff) +1 -1
@@ -1634,7 +1531,7 @@ asmlinkage long sys_futex(u32 __user *ua
                          int val3)
 {
        struct timespec t;
-       unsigned long timeout = MAX_SCHEDULE_TIMEOUT;
+       unsigned long timeout = 0;
 

This was introduced in patch-2.6.14-rt13-rf3 by David.

Quoted 5 lines

This seems to return spurious -ETIMEDOUT errors even in the non-robust code and results in userspace (glibc) retrying several mutex operations before it succeeds. I was chasing down a degradation of performance of some testcases and was able to fix those by reverting this change back.

nice catch! I've undone this change in my tree.

Ingo