[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID:
<TYZPR03MB6192BA354C0B1DE09AF1D184B3852@TYZPR03MB6192.apcprd03.prod.outlook.com>
Date: Thu, 24 Apr 2025 13:35:29 +0000
From: Adrian Huang12 <ahuang12@...ovo.com>
To: "Uladzislau Rezki (Sony)" <urezki@...il.com>,
Andrew Morton
<akpm@...ux-foundation.org>
CC: "linux-mm@...ck.org" <linux-mm@...ck.org>,
LKML
<linux-kernel@...r.kernel.org>, Baoquan He <bhe@...hat.com>,
Christoph
Hellwig <hch@...radead.org>,
Oleksiy Avramchenko
<oleksiy.avramchenko@...y.com>
Subject: RE: [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup
> -----Original Message-----
> From: owner-linux-mm@...ck.org <owner-linux-mm@...ck.org> On Behalf
> Of Uladzislau Rezki (Sony)
> Sent: Friday, April 18, 2025 12:12 AM
> To: Andrew Morton <akpm@...ux-foundation.org>
> Cc: linux-mm@...ck.org; LKML <linux-kernel@...r.kernel.org>; Baoquan He
> <bhe@...hat.com>; Christoph Hellwig <hch@...radead.org>; Uladzislau Rezki
> <urezki@...il.com>; Oleksiy Avramchenko <oleksiy.avramchenko@...y.com>
> Subject: [External] [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for
> setup
>
> The test has the initialization step during which threads are created. To
> prevent the workers from starting prematurely a write lock was previously
> used by the main setup thread, while each worker would block on a read lock.
>
> Replace this RWSEM based synchronization with a simpler SRCU based
> approach. Which does two basic steps:
>
> - Main thread wraps the setup phase in an SRCU read-side
> critical section. Pair of srcu_read_lock()/srcu_read_unlock().
> - Each worker calls synchronize_srcu() on entry, ensuring
> it waits for the initialization phase to be completed.
>
> This patch eliminates the need for down_read()/up_read() and
> down_write()/up_write() pairs thus simplifying the logic and improving clarity.
>
> Signed-off-by: Uladzislau Rezki (Sony) <urezki@...il.com>
> ---
> lib/test_vmalloc.c | 17 +++++++----------
> 1 file changed, 7 insertions(+), 10 deletions(-)
Reviewed-by: Adrian Huang <ahuang12@...ovo.com>
Tested-by: Adrian Huang <ahuang12@...ovo.com>
> diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c index
> f585949ff696e..4ab23e5e772d0 100644
> --- a/lib/test_vmalloc.c
> +++ b/lib/test_vmalloc.c
> @@ -13,9 +13,9 @@
> #include <linux/moduleparam.h>
> #include <linux/completion.h>
> #include <linux/delay.h>
> -#include <linux/rwsem.h>
> #include <linux/mm.h>
> #include <linux/rcupdate.h>
> +#include <linux/srcutree.h>
> #include <linux/slab.h>
>
> #define __param(type, name, init, msg) \
> @@ -58,10 +58,9 @@ __param(int, run_test_mask, INT_MAX, );
>
> /*
> - * Read write semaphore for synchronization of setup
> - * phase that is done in main thread and workers.
> + * This is for synchronization of setup phase.
> */
> -static DECLARE_RWSEM(prepare_for_test_rwsem);
> +DEFINE_STATIC_SRCU(prepare_for_test_srcu);
>
> /*
> * Completion tracking for worker threads.
> @@ -458,7 +457,7 @@ static int test_func(void *private)
> /*
> * Block until initialization is done.
> */
> - down_read(&prepare_for_test_rwsem);
> + synchronize_srcu(&prepare_for_test_srcu);
>
> t->start = get_cycles();
> for (i = 0; i < ARRAY_SIZE(test_case_array); i++) { @@ -487,8 +486,6 @@
> static int test_func(void *private)
> t->data[index].time = delta;
> }
> t->stop = get_cycles();
> -
> - up_read(&prepare_for_test_rwsem);
> test_report_one_done();
>
> /*
> @@ -526,7 +523,7 @@ init_test_configuration(void)
>
> static void do_concurrent_test(void)
> {
> - int i, ret;
> + int i, ret, idx;
>
> /*
> * Set some basic configurations plus sanity check.
> @@ -538,7 +535,7 @@ static void do_concurrent_test(void)
> /*
> * Put on hold all workers.
> */
> - down_write(&prepare_for_test_rwsem);
> + idx = srcu_read_lock(&prepare_for_test_srcu);
>
> for (i = 0; i < nr_threads; i++) {
> struct test_driver *t = &tdriver[i];
> @@ -555,7 +552,7 @@ static void do_concurrent_test(void)
> /*
> * Now let the workers do their job.
> */
> - up_write(&prepare_for_test_rwsem);
> + srcu_read_unlock(&prepare_for_test_srcu, idx);
>
> /*
> * Sleep quiet until all workers are done with 1 second
> --
> 2.39.5
>
Powered by blists - more mailing lists