Skip to contentSkip to conversation

mm/percpu.c: Modify calculation of size of populated bitmap of chunk for memory allocation

PATCH In review mcsmonk 3 participants 3 messages Sep 3, 2020

Subsystems (2)
lore
Share & tools
lore

mcsmonk

author
#0 lore raw

PATCH 1/1 mm/percpu.c: Modify calculation of size of populated bitmap of chunk for memory allocation

From: Sunghyun Jin <[email protected]>

Variable populated, which is a member of struct pcpu_chunk, is used as a unit of size of unsigned long.
However, size of populated is miscounted. So, I fix this minor part.

---
 mm/percpu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/percpu.c +1 -1
index f4709629e6de..1ed1a349eab8 100644
@@ -1316,7 +1316,7 @@ static struct pcpu_chunk * __init pcpu_alloc_first_chunk(unsigned long tmp_addr,
 
 	/* allocate chunk */
 	alloc_size = sizeof(struct pcpu_chunk) +
-		BITS_TO_LONGS(region_size >> PAGE_SHIFT);
+		BITS_TO_LONGS(region_size >> PAGE_SHIFT) * sizeof(unsigned long);
 	chunk = memblock_alloc(alloc_size, SMP_CACHE_BYTES);
 	if (!chunk)
 		panic("%s: Failed to allocate %zu bytes\n", __func__,
Signature · 1 line
2.17.1
#1 lore raw
Quoted 5 lines · mcsmonk

From: Sunghyun Jin <[email protected]>

Variable populated, which is a member of struct pcpu_chunk, is used as a unit of size of unsigned long.
However, size of populated is miscounted. So, I fix this minor part.

+CC folks who touched it last

Nice find! Did you observe e.g. a panic that can be used in the commit log? Or were we always lucky thanks to alignment?
Is there perhaps a commit that introduced the bug and we can use it as Fixes:? My brief look suggests 8ab16c43ea79 ("percpu: change the number of pages marked in the first_chunk pop bitmap")

Thanks!

Quoted 19 lines
---
 mm/percpu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/percpu.c +1 -1
index f4709629e6de..1ed1a349eab8 100644
@@ -1316,7 +1316,7 @@ static struct pcpu_chunk * __init pcpu_alloc_first_chunk(unsigned long tmp_addr,
 
 	/* allocate chunk */
 	alloc_size = sizeof(struct pcpu_chunk) +
-		BITS_TO_LONGS(region_size >> PAGE_SHIFT);
+		BITS_TO_LONGS(region_size >> PAGE_SHIFT) * sizeof(unsigned long);
 	chunk = memblock_alloc(alloc_size, SMP_CACHE_BYTES);
 	if (!chunk)
 		panic("%s: Failed to allocate %zu bytes\n", __func__,
maintainer
#2 lore raw

Hello Vlastimil and Sunghyun,

Quoted 9 lines · Vlastimil Babka
Quoted 5 lines · mcsmonk

From: Sunghyun Jin <[email protected]>

Variable populated, which is a member of struct pcpu_chunk, is used as a unit of size of unsigned long.
However, size of populated is miscounted. So, I fix this minor part.

+CC folks who touched it last

Thanks for CCing me.

Quoted 2 lines

Nice find! Did you observe e.g. a panic that can be used in the commit log? Or were we always lucky thanks to alignment?

Well that is indeed awkward. Luckily the first chunk is a bit special and only holds 7 dynamic pages. Additionally, the allocation rounds to SMP_CACHE_BYTES so that would give us 8 bytes to play with as struct pcpu_chunk is 120 bytes. So, while technically (wrong) the 1 byte was sufficient and the additional buffer is why at least I never got a panic report to today.

Quoted 4 lines

Is there perhaps a commit that introduced the bug and we can use it as Fixes:? My brief look suggests 8ab16c43ea79 ("percpu: change the number of pages marked in the first_chunk pop bitmap")

That is the right commit. I'll pick this up and add the fixes line.

Quoted 2 lines

Thanks!

Thanks,
Dennis

Quoted 20 lines
Quoted 19 lines
---
 mm/percpu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/percpu.c +1 -1
index f4709629e6de..1ed1a349eab8 100644
@@ -1316,7 +1316,7 @@ static struct pcpu_chunk * __init pcpu_alloc_first_chunk(unsigned long tmp_addr,
 
 	/* allocate chunk */
 	alloc_size = sizeof(struct pcpu_chunk) +
-		BITS_TO_LONGS(region_size >> PAGE_SHIFT);
+		BITS_TO_LONGS(region_size >> PAGE_SHIFT) * sizeof(unsigned long);
 	chunk = memblock_alloc(alloc_size, SMP_CACHE_BYTES);
 	if (!chunk)
 		panic("%s: Failed to allocate %zu bytes\n", __func__,