-----Original Message----- From: linux-kselftest-owner@vger.kernel.org [mailto:linux-kselftest-owner@vger.kernel.org] On Behalf Of John Hubbard Sent: Friday, August 21, 2020 3:51 PM To: Song Bao Hua (Barry Song) song.bao.hua@hisilicon.com; akpm@linux-foundation.org; linux-mm@kvack.org; linux-kselftest@vger.kernel.org Cc: Linuxarm linuxarm@huawei.com; linux-kernel@vger.kernel.org; Keith Busch keith.busch@intel.com; Ira Weiny ira.weiny@intel.com; Kirill A . Shutemov kirill.shutemov@linux.intel.com Subject: Re: [PATCH] mm/gup_benchmark: update the documentation in Kconfig
On 8/20/20 8:25 PM, Barry Song wrote:
In the beginning, mm/gup_benchmark.c supported get_user_pages_fast() only, but right now, it supports the benchmarking of a couple of get_user_pages() related calls like:
- get_user_pages_fast()
- get_user_pages()
- pin_user_pages_fast()
- pin_user_pages()
The documentation is confusing and needs update.
hmmm, it's not that confusing, given that pin_user_pages() and get_user_pages() use the same underlying get_user_pages() implementation.
get_user_pages_fast() is not get_user_pages(). It is a specific function :-)
In the beginning, gup_benchmark was only for get_user_pages_fast(): https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/mm...
Cc: John Hubbard jhubbard@nvidia.com Cc: Keith Busch keith.busch@intel.com Cc: Ira Weiny ira.weiny@intel.com Cc: Kirill A. Shutemov kirill.shutemov@linux.intel.com Signed-off-by: Barry Song song.bao.hua@hisilicon.com
mm/Kconfig | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/mm/Kconfig b/mm/Kconfig index 6c974888f86f..f7c9374da7b3 100644 --- a/mm/Kconfig +++ b/mm/Kconfig @@ -831,10 +831,10 @@ config PERCPU_STATS be used to help understand percpu memory usage.
config GUP_BENCHMARK
- bool "Enable infrastructure for get_user_pages_fast() benchmarking"
- bool "Enable infrastructure for get_user_pages() and related calls
benchmarking"
If we really want to go to the trouble of tweaking this, then I'd go with something more like:
"Enable infrastructure for get_user_pages() and pin_user_pages benchmarking"
...but I don't think it really warrants a patch just yet. *However*, my judgment is skewed right now, because I'm planning a small patchset to split up gup_benchmark a little bit, and to add some more testing and take advantage of parts of it to do a dump_page() test. At which point "related calls" would make more sense, but then it would be different enough that this patch would still need changing.
Ok. Fair enough. I was copying the "related calls" from: Documentation/core-api/pin_user_pages.rst
So I'm inclined to just recommend leaving this alone for a bit, but if others want to put it in, I'm OK with that too.
help Provides /sys/kernel/debug/gup_benchmark that helps with testing
performance of get_user_pages_fast().
performance of get_user_pages() and related calls.See tools/testing/selftests/vm/gup_benchmark.c
Thanks Barry