2024-06-11 08:57:34

by Lei Liu

[permalink] [raw]
Subject: [PATCH] binder_alloc: replace kcalloc with kvcalloc to mitigate OOM issues

In binder_alloc, there is a frequent need for order3 memory allocation,
especially on small-memory mobile devices, which can lead to OOM and
cause foreground applications to be killed, resulting in flashbacks.

We use kvcalloc to allocate memory, which can reduce system OOM
occurrences, as well as decrease the time and probability of failure for
order3 memory allocations. Additionally, it can also improve the
throughput of binder (as verified by Google's binder_benchmark testing
tool).

We have conducted multiple tests on an 8GB memory phone, and the
performance of kvcalloc is better. Below is a partial excerpt of the
test data.

throughput = (size * Iterations)/Time
Benchmark-kvcalloc Time CPU Iterations throughput(Gb/s)
----------------------------------------------------------------
BM_sendVec_binder-4096 30926 ns 20481 ns 34457 4563.66↑
BM_sendVec_binder-8192 42667 ns 30837 ns 22631 4345.11↑
BM_sendVec_binder-16384 67586 ns 52381 ns 13318 3228.51↑
BM_sendVec_binder-32768 116496 ns 94893 ns 7416 2085.97↑
BM_sendVec_binder-65536 265482 ns 209214 ns 3530 871.40↑

Benchmark-kvcalloc Time CPU Iterations throughput(Gb/s)
----------------------------------------------------------------
BM_sendVec_binder-4096 39070 ns 24207 ns 31063 3256.56
BM_sendVec_binder-8192 49476 ns 35099 ns 18817 3115.62
BM_sendVec_binder-16384 76866 ns 58924 ns 11883 2532.86
BM_sendVec_binder-32768 134022 ns 102788 ns 6535 1597.78
BM_sendVec_binder-65536 281004 ns 220028 ns 3135 731.14

Signed-off-by: Lei Liu <[email protected]>
---
drivers/android/binder_alloc.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/android/binder_alloc.c b/drivers/android/binder_alloc.c
index 2e1f261ec5c8..5dcab4a5e341 100644
--- a/drivers/android/binder_alloc.c
+++ b/drivers/android/binder_alloc.c
@@ -836,7 +836,7 @@ int binder_alloc_mmap_handler(struct binder_alloc *alloc,

alloc->buffer = vma->vm_start;

- alloc->pages = kcalloc(alloc->buffer_size / PAGE_SIZE,
+ alloc->pages = kvcalloc(alloc->buffer_size / PAGE_SIZE,
sizeof(alloc->pages[0]),
GFP_KERNEL);
if (alloc->pages == NULL) {
@@ -869,7 +869,7 @@ int binder_alloc_mmap_handler(struct binder_alloc *alloc,
return 0;

err_alloc_buf_struct_failed:
- kfree(alloc->pages);
+ kvfree(alloc->pages);
alloc->pages = NULL;
err_alloc_pages_failed:
alloc->buffer = 0;
@@ -939,7 +939,7 @@ void binder_alloc_deferred_release(struct binder_alloc *alloc)
__free_page(alloc->pages[i].page_ptr);
page_count++;
}
- kfree(alloc->pages);
+ kvfree(alloc->pages);
}
spin_unlock(&alloc->lock);
if (alloc->mm)
--
2.34.1



2024-06-12 09:59:06

by Greg Kroah-Hartman

[permalink] [raw]
Subject: Re: [PATCH] binder_alloc: replace kcalloc with kvcalloc to mitigate OOM issues

On Tue, Jun 11, 2024 at 04:56:28PM +0800, Lei Liu wrote:
> In binder_alloc, there is a frequent need for order3 memory allocation,
> especially on small-memory mobile devices, which can lead to OOM and
> cause foreground applications to be killed, resulting in flashbacks.
>
> We use kvcalloc to allocate memory, which can reduce system OOM
> occurrences, as well as decrease the time and probability of failure for
> order3 memory allocations. Additionally, it can also improve the
> throughput of binder (as verified by Google's binder_benchmark testing
> tool).
>
> We have conducted multiple tests on an 8GB memory phone, and the
> performance of kvcalloc is better. Below is a partial excerpt of the
> test data.
>
> throughput = (size * Iterations)/Time
> Benchmark-kvcalloc Time CPU Iterations throughput(Gb/s)
> ----------------------------------------------------------------
> BM_sendVec_binder-4096 30926 ns 20481 ns 34457 4563.66↑
> BM_sendVec_binder-8192 42667 ns 30837 ns 22631 4345.11↑
> BM_sendVec_binder-16384 67586 ns 52381 ns 13318 3228.51↑
> BM_sendVec_binder-32768 116496 ns 94893 ns 7416 2085.97↑
> BM_sendVec_binder-65536 265482 ns 209214 ns 3530 871.40↑
>
> Benchmark-kvcalloc Time CPU Iterations throughput(Gb/s)

Both benchmarks are the same? Or is this labeled incorrectly?

> ----------------------------------------------------------------
> BM_sendVec_binder-4096 39070 ns 24207 ns 31063 3256.56
> BM_sendVec_binder-8192 49476 ns 35099 ns 18817 3115.62
> BM_sendVec_binder-16384 76866 ns 58924 ns 11883 2532.86
> BM_sendVec_binder-32768 134022 ns 102788 ns 6535 1597.78
> BM_sendVec_binder-65536 281004 ns 220028 ns 3135 731.14
>
> Signed-off-by: Lei Liu <[email protected]>
> ---
> drivers/android/binder_alloc.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/android/binder_alloc.c b/drivers/android/binder_alloc.c
> index 2e1f261ec5c8..5dcab4a5e341 100644
> --- a/drivers/android/binder_alloc.c
> +++ b/drivers/android/binder_alloc.c
> @@ -836,7 +836,7 @@ int binder_alloc_mmap_handler(struct binder_alloc *alloc,
>
> alloc->buffer = vma->vm_start;
>
> - alloc->pages = kcalloc(alloc->buffer_size / PAGE_SIZE,
> + alloc->pages = kvcalloc(alloc->buffer_size / PAGE_SIZE,
> sizeof(alloc->pages[0]),
> GFP_KERNEL);

Nit, update the indentation please.

thanks,

greg k-h

2024-06-13 12:07:08

by Lei Liu

[permalink] [raw]
Subject: Re: [PATCH] binder_alloc: replace kcalloc with kvcalloc to mitigate OOM issues

On 2024/6/12 17:58, Greg Kroah-Hartman wrote:
> On Tue, Jun 11, 2024 at 04:56:28PM +0800, Lei Liu wrote:
>
>> In binder_alloc, there is a frequent need for order3 memory
>> allocation, especially on small-memory mobile devices, which can lead
>> to OOM and cause foreground applications to be killed, resulting in
>> flashbacks. We use kvcalloc to allocate memory, which can reduce
>> system OOM occurrences, as well as decrease the time and probability
>> of failure for order3 memory allocations. Additionally, it can also
>> improve the throughput of binder (as verified by Google's
>> binder_benchmark testing tool). We have conducted multiple tests on
>> an 8GB memory phone, and the performance of kvcalloc is better. Below
>> is a partial excerpt of the test data. throughput = (size *
>> Iterations)/Time Benchmark-kvcalloc Time CPU Iterations
>> throughput(Gb/s)
>> ----------------------------------------------------------------
>> BM_sendVec_binder-4096 30926 ns 20481 ns 34457 4563.66↑
>> BM_sendVec_binder-8192 42667 ns 30837 ns 22631 4345.11↑
>> BM_sendVec_binder-16384 67586 ns 52381 ns 13318 3228.51↑
>> BM_sendVec_binder-32768 116496 ns 94893 ns 7416 2085.97↑
>> BM_sendVec_binder-65536 265482 ns 209214 ns 3530 871.40↑
>> Benchmark-kvcalloc Time CPU Iterations throughput(Gb/s)
> Both benchmarks are the same? Or is this labeled incorrectly?
I'm really sorry, I got the title of the table wrong, here are the
updated data:
throughput = (size * Iterations)/Time
kvcalloc->kvmalloc:
Benchmark-kvcalloc    Time    CPU    Iterations    throughput(Gb/s)
----------------------------------------------------------------
BM_sendVec_binder-4096    30926 ns    20481 ns    34457    4563.66↑
BM_sendVec_binder-8192    42667 ns    30837 ns    22631    4345.11↑
BM_sendVec_binder-16384    67586 ns    52381 ns    13318    3228.51↑
BM_sendVec_binder-32768    116496 ns    94893 ns    7416    2085.97↑
BM_sendVec_binder-65536    265482 ns    209214 ns    3530    871.40↑

kcalloc->kmalloc
Benchmark-kcalloc    Time    CPU    Iterations    throughput(Gb/s)
----------------------------------------------------------------
BM_sendVec_binder-4096    39070 ns    24207 ns    31063    3256.56
BM_sendVec_binder-8192    49476 ns    35099 ns    18817    3115.62
BM_sendVec_binder-16384    76866 ns    58924 ns    11883    2532.86
BM_sendVec_binder-32768    134022 ns    102788 ns    6535    1597.78
BM_sendVec_binder-65536    281004 ns    220028 ns    3135    731.14
>> ----------------------------------------------------------------
>> BM_sendVec_binder-4096 39070 ns 24207 ns 31063 3256.56
>> BM_sendVec_binder-8192 49476 ns 35099 ns 18817 3115.62
>> BM_sendVec_binder-16384 76866 ns 58924 ns 11883 2532.86
>> BM_sendVec_binder-32768 134022 ns 102788 ns 6535 1597.78
>> BM_sendVec_binder-65536 281004 ns 220028 ns 3135 731.14
>> Signed-off-by: Lei Liu <[email protected]> ---
>> drivers/android/binder_alloc.c | 6 +++--- 1 file changed, 3
>> insertions(+), 3 deletions(-) diff --git
>> a/drivers/android/binder_alloc.c b/drivers/android/binder_alloc.c
>> index 2e1f261ec5c8..5dcab4a5e341 100644 ---
>> a/drivers/android/binder_alloc.c +++ b/drivers/android/binder_alloc.c
>> @@ -836,7 +836,7 @@ int binder_alloc_mmap_handler(struct binder_alloc
>> *alloc, alloc->buffer = vma->vm_start; - alloc->pages =
>> kcalloc(alloc->buffer_size / PAGE_SIZE, + alloc->pages =
>> kvcalloc(alloc->buffer_size / PAGE_SIZE, sizeof(alloc->pages[0]),
>> GFP_KERNEL);
> Nit, update the indentation please.
> thanks, greg k-h
The previous incorrect table title has been updated. Please help review
it again.

Thank you!
Lei Liu

2024-06-13 13:42:43

by Greg Kroah-Hartman

[permalink] [raw]
Subject: Re: [PATCH] binder_alloc: replace kcalloc with kvcalloc to mitigate OOM issues

On Thu, Jun 13, 2024 at 08:01:39PM +0800, Lei Liu wrote:
> On 2024/6/12 17:58, Greg Kroah-Hartman wrote:
> > On Tue, Jun 11, 2024 at 04:56:28PM +0800, Lei Liu wrote:
> >
> > > In binder_alloc, there is a frequent need for order3 memory
> > > allocation, especially on small-memory mobile devices, which can
> > > lead to OOM and cause foreground applications to be killed,
> > > resulting in flashbacks. We use kvcalloc to allocate memory, which
> > > can reduce system OOM occurrences, as well as decrease the time and
> > > probability of failure for order3 memory allocations. Additionally,
> > > it can also improve the throughput of binder (as verified by
> > > Google's binder_benchmark testing tool). We have conducted multiple
> > > tests on an 8GB memory phone, and the performance of kvcalloc is
> > > better. Below is a partial excerpt of the test data. throughput =
> > > (size * Iterations)/Time Benchmark-kvcalloc Time CPU Iterations
> > > throughput(Gb/s)
> > > ----------------------------------------------------------------
> > > BM_sendVec_binder-4096 30926 ns 20481 ns 34457 4563.66↑
> > > BM_sendVec_binder-8192 42667 ns 30837 ns 22631 4345.11↑
> > > BM_sendVec_binder-16384 67586 ns 52381 ns 13318 3228.51↑
> > > BM_sendVec_binder-32768 116496 ns 94893 ns 7416 2085.97↑
> > > BM_sendVec_binder-65536 265482 ns 209214 ns 3530 871.40↑
> > > Benchmark-kvcalloc Time CPU Iterations throughput(Gb/s)
> > Both benchmarks are the same? Or is this labeled incorrectly?
> I'm really sorry, I got the title of the table wrong, here are the updated
> data:
> throughput = (size * Iterations)/Time
> kvcalloc->kvmalloc:
> Benchmark-kvcalloc    Time    CPU    Iterations    throughput(Gb/s)
> ----------------------------------------------------------------
> BM_sendVec_binder-4096    30926 ns    20481 ns    34457    4563.66↑
> BM_sendVec_binder-8192    42667 ns    30837 ns    22631    4345.11↑
> BM_sendVec_binder-16384    67586 ns    52381 ns    13318    3228.51↑
> BM_sendVec_binder-32768    116496 ns    94893 ns    7416    2085.97↑
> BM_sendVec_binder-65536    265482 ns    209214 ns    3530    871.40↑
>
> kcalloc->kmalloc
> Benchmark-kcalloc    Time    CPU    Iterations    throughput(Gb/s)
> ----------------------------------------------------------------
> BM_sendVec_binder-4096    39070 ns    24207 ns    31063    3256.56
> BM_sendVec_binder-8192    49476 ns    35099 ns    18817    3115.62
> BM_sendVec_binder-16384    76866 ns    58924 ns    11883    2532.86
> BM_sendVec_binder-32768    134022 ns    102788 ns    6535    1597.78
> BM_sendVec_binder-65536    281004 ns    220028 ns    3135    731.14

Great, can you please resend this as a new version?

thanks,

greg k-h