Skip to contentSkip to conversation

cpuset and sched domains: sched_load_balance flag fixes

PATCH In review Paul Jackson 3 participants 7 messages Oct 3, 2007 – Oct 6, 2007

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 cpuset and sched domains: sched_load_balance flag fixes

From: Paul Jackson <[email protected]>

Thanks to Randy Dunlap for the review that caught some of the following.

Some bug fixes and coding style fixes:
1) only one statement per line, please.
2) don't need to guard kfree() calls with a NULL check
3) use kfifo_free, not kfree, if it came from kfifo_alloc
4) a pair of curly brackets got lost along the way
5) missing .read, .write callbacks for sched_load_balance

Without (3), one kfifo buffer memory was leaked each time one rebuilt scheduler domains

Without (4), the current task was summarily killed each time one tried to rebuild scheduler domains

Without (5), every read or write system call on a per-cpuset special file 'sched_load_balance' failed, EINVAL.

---

Andrew,

These fixes go right after the patch they fix:
[PATCH] cpuset and sched domains: sched_load_balance flag

 kernel/cpuset.c |   14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
kernel/cpuset.c +9 -5
@@ -579,7 +579,9 @@ static void rebuild_sched_domains(void)
 	int ndoms;		/* number of sched domains in result */
 	int nslot;		/* next empty doms[] cpumask_t slot */
 
-	q = NULL; csa = NULL; doms = NULL;
+	q = NULL;
+	csa = NULL;
+	doms = NULL;
 
 	/* Special case for the 99% of systems with one, full, sched domain */
 	if (is_sched_load_balance(&top_cpuset)) {
@@ -604,9 +606,10 @@ static void rebuild_sched_domains(void)
 		struct cpuset *child;   /* scans child cpusets of cp */
 		if (is_sched_load_balance(cp))
 			csa[csn++] = cp;
-		list_for_each_entry(cont, &cp->css.cgroup->children, sibling)
+		list_for_each_entry(cont, &cp->css.cgroup->children, sibling) {
 			child = cgroup_cs(cont);
 			__kfifo_put(q, (void *)&child, sizeof(cp));
+		}
   	}
 
 	for (i = 0; i < csn; i++)
@@ -668,9 +671,8 @@ rebuild:
 
 done:
 	if (q && !IS_ERR(q))
-		kfree(q);
-	if (csa)
-		kfree(csa);
+		kfifo_free(q);
+	kfree(csa);
 	/* Don't kfree(doms) -- partition_sched_domains() does that. */
 }
 
@@ -1358,6 +1360,8 @@ static struct cftype cft_mem_exclusive =
 
 static struct cftype cft_sched_load_balance = {
 	.name = "sched_load_balance",
+	.read = cpuset_common_file_read,
+	.write = cpuset_common_file_write,
 	.private = FILE_SCHED_LOAD_BALANCE,
 };
 
Signature · 3 lines
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <[email protected]> 1.650.933.1373
#1 lore raw
Quoted 21 lines · On Wed, 03 Oct 2007 04:44:01 -0700 Paul Jackson <[email protected]> wrote:

From: Paul Jackson <[email protected]>

Thanks to Randy Dunlap for the review that caught some of the following.

Some bug fixes and coding style fixes:
1) only one statement per line, please.
2) don't need to guard kfree() calls with a NULL check
3) use kfifo_free, not kfree, if it came from kfifo_alloc
4) a pair of curly brackets got lost along the way
5) missing .read, .write callbacks for sched_load_balance

Without (3), one kfifo buffer memory was leaked each time one rebuilt scheduler domains

Without (4), the current task was summarily killed each time one tried to rebuild scheduler domains

Without (5), every read or write system call on a per-cpuset special file 'sched_load_balance' failed, EINVAL.

Quoted 8 lines

---

Andrew,

These fixes go right after the patch they fix:
[PATCH] cpuset and sched domains: sched_load_balance flag

I'm getting 100% rejects from this - probably whatever patch it is patching got lost or unrecognisably mangled.

#2 lore raw
Quoted 2 lines · Andrew

I'm getting 100% rejects from this - probably whatever patch it is patching got lost or unrecognisably mangled.

Yup - so far as I can tell, you didn't pick up the base patch that this current patch fixes. So, no, I wouldn't expect this patch to make any sense.

As I stated in this current patch in the diffstat section after the '---' marker, this current patch applies to the base patch of Subject:

[PATCH] cpuset and sched domains: sched_load_balance flag

You probably didn't pick up that base patch because Nick and I are still haggling over it. Well ... we've agreed, but I can't get the scheduler to work as I thought Nick wanted, and I'm still waiting to hear from Nick again.

==> So yeah - throw this current patch out.

Nick ... you there ?

Signature · 3 lines
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <[email protected]> 1.925.600.0401
#3 lore raw
Quoted 7 lines · Paul Jackson

Yup - so far as I can tell, you didn't pick up the base patch that this current patch fixes. So, no, I wouldn't expect this patch to make any sense.

As I stated in this current patch in the diffstat section after the '---' marker, this current patch applies to the base patch of Subject:

Quoted 6 lines

[PATCH] cpuset and sched domains: sched_load_balance flag

You probably didn't pick up that base patch because Nick and I are still haggling over it. Well ... we've agreed, but I can't get the scheduler to work as I thought Nick wanted, and I'm still waiting to hear from Nick again.

please resend the base patch to Andrew (or better yet, a combo patch that Andrew can apply and which just does it all).

We've agreed on the external API and that is what matters for now. The internal interfacing between the scheduler and cpusets can/could be improved later on - and your original patch is reasonable to begin with. So lets just move on and do this.

Ingo

#4 lore raw
Quoted 2 lines · Ingo

please resend the base patch to Andrew (or better yet, a combo patch that Andrew can apply and which just does it all).

I'm reluctant to right now, because I will be off the air for three days, starting in one more day, and don't like firing off stuff just before I vanish.

Though I'm aware that time is short for 2.6.24 ... not sure how best to handle this.

My plan had been to send Andrew something on Wednesday next week (five days from now), either what we have now, or with the improved cpuset-to-sched API, if Nick and I can work through that (well, if Nick can figure out what I broke, while I'm offline.)

Andrew or Ingo - you got any better plan?

Signature · 3 lines
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <[email protected]> 1.925.600.0401
#5 lore raw
Quoted 7 lines · Paul Jackson

My plan had been to send Andrew something on Wednesday next week (five days from now), either what we have now, or with the improved cpuset-to-sched API, if Nick and I can work through that (well, if Nick can figure out what I broke, while I'm offline.)

Andrew or Ingo - you got any better plan?

i'd suggest we go with what we have now (i've run it for some reasonable time on various systems and it caused no breakages) - could you resend that? We cannot merge it via sched.git because it has some containers (or other?) dependencies, right?

Ingo

#6 lore raw
Quoted 1 line

i'd suggest we go with what we have now

Ok - I'll try sending this against 2.6.23-rc8-mm2
in a couple of hours, if it goes well.

Quoted 2 lines

We cannot merge it via sched.git because it has some containers (or other?) dependencies, right?

This sched_load_balance flag patch collides with the cgroup (aka container) patches if I'm not careful, yes.

Signature · 3 lines
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <[email protected]> 1.925.600.0401