* [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup
@ 2025-04-17 16:12 Uladzislau Rezki (Sony)
2025-04-17 16:12 ` [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution Uladzislau Rezki (Sony)
` (4 more replies)
0 siblings, 5 replies; 18+ messages in thread
From: Uladzislau Rezki (Sony) @ 2025-04-17 16:12 UTC (permalink / raw)
To: Andrew Morton
Cc: linux-mm, LKML, Baoquan He, Christoph Hellwig, Uladzislau Rezki,
Oleksiy Avramchenko
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@gmail.com>
---
lib/test_vmalloc.c | 17 +++++++----------
1 file changed, 7 insertions(+), 10 deletions(-)
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
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution 2025-04-17 16:12 [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup Uladzislau Rezki (Sony) @ 2025-04-17 16:12 ` Uladzislau Rezki (Sony) 2025-04-18 15:01 ` Baoquan He ` (2 more replies) 2025-04-17 16:12 ` [PATCH 3/4] MAINTAINERS: Add test_vmalloc.c to VMALLOC section Uladzislau Rezki (Sony) ` (3 subsequent siblings) 4 siblings, 3 replies; 18+ messages in thread From: Uladzislau Rezki (Sony) @ 2025-04-17 16:12 UTC (permalink / raw) To: Andrew Morton Cc: linux-mm, LKML, Baoquan He, Christoph Hellwig, Uladzislau Rezki, Oleksiy Avramchenko This patch removes the dependency on module loading ("m") for the vmalloc test suite, enabling it to be built directly into the kernel, so both ("=m") and ("=y") are supported. Motivation: - Faster debugging/testing of vmalloc code; - It allows to configure the test via kernel-boot parameters. Configuration example: test_vmalloc.nr_threads=64 test_vmalloc.run_test_mask=7 test_vmalloc.sequential_test_order=1 Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> --- lib/Kconfig.debug | 3 +-- lib/test_vmalloc.c | 5 +++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug index f9051ab610d54..166b9d830a85a 100644 --- a/lib/Kconfig.debug +++ b/lib/Kconfig.debug @@ -2574,8 +2574,7 @@ config TEST_BITOPS config TEST_VMALLOC tristate "Test module for stress/performance analysis of vmalloc allocator" default n - depends on MMU - depends on m + depends on MMU help This builds the "test_vmalloc" module that should be used for stress and performance analysis. So, any new change for vmalloc diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c index 4ab23e5e772d0..6d65ef725d42c 100644 --- a/lib/test_vmalloc.c +++ b/lib/test_vmalloc.c @@ -591,10 +591,11 @@ static void do_concurrent_test(void) kvfree(tdriver); } -static int vmalloc_test_init(void) +static int __init vmalloc_test_init(void) { do_concurrent_test(); - return -EAGAIN; /* Fail will directly unload the module */ + /* Fail will directly unload the module */ + return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN; } module_init(vmalloc_test_init) -- 2.39.5 ^ permalink raw reply related [flat|nested] 18+ messages in thread
* Re: [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution 2025-04-17 16:12 ` [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution Uladzislau Rezki (Sony) @ 2025-04-18 15:01 ` Baoquan He 2025-04-24 13:37 ` Adrian Huang12 2025-06-13 13:59 ` Dev Jain 2 siblings, 0 replies; 18+ messages in thread From: Baoquan He @ 2025-04-18 15:01 UTC (permalink / raw) To: Uladzislau Rezki (Sony) Cc: Andrew Morton, linux-mm, LKML, Christoph Hellwig, Oleksiy Avramchenko On 04/17/25 at 06:12pm, Uladzislau Rezki (Sony) wrote: > This patch removes the dependency on module loading ("m") > for the vmalloc test suite, enabling it to be built directly > into the kernel, so both ("=m") and ("=y") are supported. > > Motivation: > - Faster debugging/testing of vmalloc code; > - It allows to configure the test via kernel-boot parameters. > > Configuration example: > test_vmalloc.nr_threads=64 > test_vmalloc.run_test_mask=7 > test_vmalloc.sequential_test_order=1 > > Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> > --- > lib/Kconfig.debug | 3 +-- > lib/test_vmalloc.c | 5 +++-- > 2 files changed, 4 insertions(+), 4 deletions(-) Reviewed-by: Baoquan He <bhe@redhat.com> > > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug > index f9051ab610d54..166b9d830a85a 100644 > --- a/lib/Kconfig.debug > +++ b/lib/Kconfig.debug > @@ -2574,8 +2574,7 @@ config TEST_BITOPS > config TEST_VMALLOC > tristate "Test module for stress/performance analysis of vmalloc allocator" > default n > - depends on MMU > - depends on m > + depends on MMU > help > This builds the "test_vmalloc" module that should be used for > stress and performance analysis. So, any new change for vmalloc > diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c > index 4ab23e5e772d0..6d65ef725d42c 100644 > --- a/lib/test_vmalloc.c > +++ b/lib/test_vmalloc.c > @@ -591,10 +591,11 @@ static void do_concurrent_test(void) > kvfree(tdriver); > } > > -static int vmalloc_test_init(void) > +static int __init vmalloc_test_init(void) > { > do_concurrent_test(); > - return -EAGAIN; /* Fail will directly unload the module */ > + /* Fail will directly unload the module */ > + return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN; > } > > module_init(vmalloc_test_init) > -- > 2.39.5 > ^ permalink raw reply [flat|nested] 18+ messages in thread
* RE: [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution 2025-04-17 16:12 ` [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution Uladzislau Rezki (Sony) 2025-04-18 15:01 ` Baoquan He @ 2025-04-24 13:37 ` Adrian Huang12 2025-06-13 13:59 ` Dev Jain 2 siblings, 0 replies; 18+ messages in thread From: Adrian Huang12 @ 2025-04-24 13:37 UTC (permalink / raw) To: Uladzislau Rezki (Sony), Andrew Morton Cc: linux-mm@kvack.org, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko > -----Original Message----- > From: owner-linux-mm@kvack.org <owner-linux-mm@kvack.org> On Behalf > Of Uladzislau Rezki (Sony) > Sent: Friday, April 18, 2025 12:12 AM > To: Andrew Morton <akpm@linux-foundation.org> > Cc: linux-mm@kvack.org; LKML <linux-kernel@vger.kernel.org>; Baoquan He > <bhe@redhat.com>; Christoph Hellwig <hch@infradead.org>; Uladzislau Rezki > <urezki@gmail.com>; Oleksiy Avramchenko <oleksiy.avramchenko@sony.com> > Subject: [External] [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution > > This patch removes the dependency on module loading ("m") for the vmalloc > test suite, enabling it to be built directly into the kernel, so both ("=m") and > ("=y") are supported. > > Motivation: > - Faster debugging/testing of vmalloc code; > - It allows to configure the test via kernel-boot parameters. > > Configuration example: > test_vmalloc.nr_threads=64 > test_vmalloc.run_test_mask=7 > test_vmalloc.sequential_test_order=1 > > Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> > --- > lib/Kconfig.debug | 3 +-- > lib/test_vmalloc.c | 5 +++-- > 2 files changed, 4 insertions(+), 4 deletions(-) Reviewed-by: Adrian Huang <ahuang12@lenovo.com> Tested-by: Adrian Huang <ahuang12@lenovo.com> > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug index > f9051ab610d54..166b9d830a85a 100644 > --- a/lib/Kconfig.debug > +++ b/lib/Kconfig.debug > @@ -2574,8 +2574,7 @@ config TEST_BITOPS config TEST_VMALLOC > tristate "Test module for stress/performance analysis of vmalloc allocator" > default n > - depends on MMU > - depends on m > + depends on MMU > help > This builds the "test_vmalloc" module that should be used for > stress and performance analysis. So, any new change for vmalloc diff > --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c index > 4ab23e5e772d0..6d65ef725d42c 100644 > --- a/lib/test_vmalloc.c > +++ b/lib/test_vmalloc.c > @@ -591,10 +591,11 @@ static void do_concurrent_test(void) > kvfree(tdriver); > } > > -static int vmalloc_test_init(void) > +static int __init vmalloc_test_init(void) > { > do_concurrent_test(); > - return -EAGAIN; /* Fail will directly unload the module */ > + /* Fail will directly unload the module */ > + return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN; > } > > module_init(vmalloc_test_init) > -- > 2.39.5 > ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution 2025-04-17 16:12 ` [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution Uladzislau Rezki (Sony) 2025-04-18 15:01 ` Baoquan He 2025-04-24 13:37 ` Adrian Huang12 @ 2025-06-13 13:59 ` Dev Jain 2025-06-13 14:19 ` Uladzislau Rezki 2 siblings, 1 reply; 18+ messages in thread From: Dev Jain @ 2025-06-13 13:59 UTC (permalink / raw) To: Uladzislau Rezki (Sony), Andrew Morton Cc: linux-mm, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko, Ryan Roberts On 17/04/25 9:42 pm, Uladzislau Rezki (Sony) wrote: > This patch removes the dependency on module loading ("m") > for the vmalloc test suite, enabling it to be built directly > into the kernel, so both ("=m") and ("=y") are supported. > > Motivation: > - Faster debugging/testing of vmalloc code; > - It allows to configure the test via kernel-boot parameters. > > Configuration example: > test_vmalloc.nr_threads=64 > test_vmalloc.run_test_mask=7 > test_vmalloc.sequential_test_order=1 > > Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> > --- > lib/Kconfig.debug | 3 +-- > lib/test_vmalloc.c | 5 +++-- > 2 files changed, 4 insertions(+), 4 deletions(-) > > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug > index f9051ab610d54..166b9d830a85a 100644 > --- a/lib/Kconfig.debug > +++ b/lib/Kconfig.debug > @@ -2574,8 +2574,7 @@ config TEST_BITOPS > config TEST_VMALLOC > tristate "Test module for stress/performance analysis of vmalloc allocator" > default n > - depends on MMU > - depends on m > + depends on MMU > help > This builds the "test_vmalloc" module that should be used for > stress and performance analysis. So, any new change for vmalloc > diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c > index 4ab23e5e772d0..6d65ef725d42c 100644 > --- a/lib/test_vmalloc.c > +++ b/lib/test_vmalloc.c > @@ -591,10 +591,11 @@ static void do_concurrent_test(void) > kvfree(tdriver); > } > > -static int vmalloc_test_init(void) > +static int __init vmalloc_test_init(void) > { > do_concurrent_test(); > - return -EAGAIN; /* Fail will directly unload the module */ > + /* Fail will directly unload the module */ > + return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN; > } > > module_init(vmalloc_test_init) Hi Uladzislau, Post this patch I am seeing the following problem: while booting the kernel, this test is executed and a particular testcase triggers a warning splat from the kernel, then the boot gets stuck for 3-4 seconds, then proceeds. I am not an expert on modules so I don't know what happened but I am guessing that when the test fails, the delay is due to module unloading time...do you suggest a solution? I reverted your patch and then proceeded with my work, so apologies for not having the dmesg log, if you need it I'll replicate, although there is some subtest which has consistently been spitting out a warning on my Apple M3 Linux VM and I have been just ignoring it for long : ) ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution 2025-06-13 13:59 ` Dev Jain @ 2025-06-13 14:19 ` Uladzislau Rezki 2025-06-13 14:23 ` Dev Jain 0 siblings, 1 reply; 18+ messages in thread From: Uladzislau Rezki @ 2025-06-13 14:19 UTC (permalink / raw) To: Dev Jain Cc: Uladzislau Rezki (Sony), Andrew Morton, linux-mm, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko, Ryan Roberts Hello, Dev! > > On 17/04/25 9:42 pm, Uladzislau Rezki (Sony) wrote: > > This patch removes the dependency on module loading ("m") > > for the vmalloc test suite, enabling it to be built directly > > into the kernel, so both ("=m") and ("=y") are supported. > > > > Motivation: > > - Faster debugging/testing of vmalloc code; > > - It allows to configure the test via kernel-boot parameters. > > > > Configuration example: > > test_vmalloc.nr_threads=64 > > test_vmalloc.run_test_mask=7 > > test_vmalloc.sequential_test_order=1 > > > > Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> > > --- > > lib/Kconfig.debug | 3 +-- > > lib/test_vmalloc.c | 5 +++-- > > 2 files changed, 4 insertions(+), 4 deletions(-) > > > > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug > > index f9051ab610d54..166b9d830a85a 100644 > > --- a/lib/Kconfig.debug > > +++ b/lib/Kconfig.debug > > @@ -2574,8 +2574,7 @@ config TEST_BITOPS > > config TEST_VMALLOC > > tristate "Test module for stress/performance analysis of vmalloc allocator" > > default n > > - depends on MMU > > - depends on m > > + depends on MMU > > help > > This builds the "test_vmalloc" module that should be used for > > stress and performance analysis. So, any new change for vmalloc > > diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c > > index 4ab23e5e772d0..6d65ef725d42c 100644 > > --- a/lib/test_vmalloc.c > > +++ b/lib/test_vmalloc.c > > @@ -591,10 +591,11 @@ static void do_concurrent_test(void) > > kvfree(tdriver); > > } > > -static int vmalloc_test_init(void) > > +static int __init vmalloc_test_init(void) > > { > > do_concurrent_test(); > > - return -EAGAIN; /* Fail will directly unload the module */ > > + /* Fail will directly unload the module */ > > + return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN; > > } > > module_init(vmalloc_test_init) > > Hi Uladzislau, > > Post this patch I am seeing the following problem: while booting the kernel, this test is > executed and a particular testcase triggers a warning splat from the kernel, then the boot > gets stuck for 3-4 seconds, then proceeds. I am not an expert on modules so I don't know > what happened but I am guessing that when the test fails, the delay is due to module > unloading time...do you suggest a solution? I reverted your patch and then proceeded with > my work, so apologies for not having the dmesg log, if you need it I'll replicate, although > there is some subtest which has consistently been spitting out a warning on my Apple M3 Linux > VM and I have been just ignoring it for long : ) > This is because CONFIG_TEST_VMALLOC=y is built-in in your case. Therefore the tests are triggered including the ones which generates kernel warnings, which is expectable. It takes some time to execute, therefore you see a delay. By default the CONFIG_TEST_VMALLOC is not set. This is just for debugging purpose. Please unset it in your kernel configuration. It is located under: Kernel Hacking Kernel Testing and Coverage Runtime Testing Test Module for stress/performance analysis of vmalloc allocator -- Uladzislau Rezki ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution 2025-06-13 14:19 ` Uladzislau Rezki @ 2025-06-13 14:23 ` Dev Jain 2025-06-13 14:37 ` Uladzislau Rezki 0 siblings, 1 reply; 18+ messages in thread From: Dev Jain @ 2025-06-13 14:23 UTC (permalink / raw) To: Uladzislau Rezki Cc: Andrew Morton, linux-mm, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko, Ryan Roberts On 13/06/25 7:49 pm, Uladzislau Rezki wrote: > Hello, Dev! > >> On 17/04/25 9:42 pm, Uladzislau Rezki (Sony) wrote: >>> This patch removes the dependency on module loading ("m") >>> for the vmalloc test suite, enabling it to be built directly >>> into the kernel, so both ("=m") and ("=y") are supported. >>> >>> Motivation: >>> - Faster debugging/testing of vmalloc code; >>> - It allows to configure the test via kernel-boot parameters. >>> >>> Configuration example: >>> test_vmalloc.nr_threads=64 >>> test_vmalloc.run_test_mask=7 >>> test_vmalloc.sequential_test_order=1 >>> >>> Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> >>> --- >>> lib/Kconfig.debug | 3 +-- >>> lib/test_vmalloc.c | 5 +++-- >>> 2 files changed, 4 insertions(+), 4 deletions(-) >>> >>> diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug >>> index f9051ab610d54..166b9d830a85a 100644 >>> --- a/lib/Kconfig.debug >>> +++ b/lib/Kconfig.debug >>> @@ -2574,8 +2574,7 @@ config TEST_BITOPS >>> config TEST_VMALLOC >>> tristate "Test module for stress/performance analysis of vmalloc allocator" >>> default n >>> - depends on MMU >>> - depends on m >>> + depends on MMU >>> help >>> This builds the "test_vmalloc" module that should be used for >>> stress and performance analysis. So, any new change for vmalloc >>> diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c >>> index 4ab23e5e772d0..6d65ef725d42c 100644 >>> --- a/lib/test_vmalloc.c >>> +++ b/lib/test_vmalloc.c >>> @@ -591,10 +591,11 @@ static void do_concurrent_test(void) >>> kvfree(tdriver); >>> } >>> -static int vmalloc_test_init(void) >>> +static int __init vmalloc_test_init(void) >>> { >>> do_concurrent_test(); >>> - return -EAGAIN; /* Fail will directly unload the module */ >>> + /* Fail will directly unload the module */ >>> + return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN; >>> } >>> module_init(vmalloc_test_init) >> Hi Uladzislau, >> >> Post this patch I am seeing the following problem: while booting the kernel, this test is >> executed and a particular testcase triggers a warning splat from the kernel, then the boot >> gets stuck for 3-4 seconds, then proceeds. I am not an expert on modules so I don't know >> what happened but I am guessing that when the test fails, the delay is due to module >> unloading time...do you suggest a solution? I reverted your patch and then proceeded with >> my work, so apologies for not having the dmesg log, if you need it I'll replicate, although >> there is some subtest which has consistently been spitting out a warning on my Apple M3 Linux >> VM and I have been just ignoring it for long : ) >> > This is because CONFIG_TEST_VMALLOC=y is built-in in your case. Actually, I have a usecase for test_vmalloc :) I wanted to test my patches through the test, but the problem is that the moment it fails, it gets unloaded and then I cannot use the test at runtime. Currently it is set to =m in my .config. > Therefore the tests are triggered including the ones which generates > kernel warnings, which is expectable. It takes some time to execute, > therefore you see a delay. > > By default the CONFIG_TEST_VMALLOC is not set. This is just for > debugging purpose. Please unset it in your kernel configuration. > It is located under: > > Kernel Hacking > Kernel Testing and Coverage > Runtime Testing > Test Module for stress/performance analysis of vmalloc allocator > > -- > Uladzislau Rezki ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution 2025-06-13 14:23 ` Dev Jain @ 2025-06-13 14:37 ` Uladzislau Rezki 2025-06-14 10:27 ` Dev Jain 0 siblings, 1 reply; 18+ messages in thread From: Uladzislau Rezki @ 2025-06-13 14:37 UTC (permalink / raw) To: Dev Jain Cc: Uladzislau Rezki, Andrew Morton, linux-mm, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko, Ryan Roberts On Fri, Jun 13, 2025 at 07:53:37PM +0530, Dev Jain wrote: > > On 13/06/25 7:49 pm, Uladzislau Rezki wrote: > > Hello, Dev! > > > > > On 17/04/25 9:42 pm, Uladzislau Rezki (Sony) wrote: > > > > This patch removes the dependency on module loading ("m") > > > > for the vmalloc test suite, enabling it to be built directly > > > > into the kernel, so both ("=m") and ("=y") are supported. > > > > > > > > Motivation: > > > > - Faster debugging/testing of vmalloc code; > > > > - It allows to configure the test via kernel-boot parameters. > > > > > > > > Configuration example: > > > > test_vmalloc.nr_threads=64 > > > > test_vmalloc.run_test_mask=7 > > > > test_vmalloc.sequential_test_order=1 > > > > > > > > Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> > > > > --- > > > > lib/Kconfig.debug | 3 +-- > > > > lib/test_vmalloc.c | 5 +++-- > > > > 2 files changed, 4 insertions(+), 4 deletions(-) > > > > > > > > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug > > > > index f9051ab610d54..166b9d830a85a 100644 > > > > --- a/lib/Kconfig.debug > > > > +++ b/lib/Kconfig.debug > > > > @@ -2574,8 +2574,7 @@ config TEST_BITOPS > > > > config TEST_VMALLOC > > > > tristate "Test module for stress/performance analysis of vmalloc allocator" > > > > default n > > > > - depends on MMU > > > > - depends on m > > > > + depends on MMU > > > > help > > > > This builds the "test_vmalloc" module that should be used for > > > > stress and performance analysis. So, any new change for vmalloc > > > > diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c > > > > index 4ab23e5e772d0..6d65ef725d42c 100644 > > > > --- a/lib/test_vmalloc.c > > > > +++ b/lib/test_vmalloc.c > > > > @@ -591,10 +591,11 @@ static void do_concurrent_test(void) > > > > kvfree(tdriver); > > > > } > > > > -static int vmalloc_test_init(void) > > > > +static int __init vmalloc_test_init(void) > > > > { > > > > do_concurrent_test(); > > > > - return -EAGAIN; /* Fail will directly unload the module */ > > > > + /* Fail will directly unload the module */ > > > > + return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN; > > > > } > > > > module_init(vmalloc_test_init) > > > Hi Uladzislau, > > > > > > Post this patch I am seeing the following problem: while booting the kernel, this test is > > > executed and a particular testcase triggers a warning splat from the kernel, then the boot > > > gets stuck for 3-4 seconds, then proceeds. I am not an expert on modules so I don't know > > > what happened but I am guessing that when the test fails, the delay is due to module > > > unloading time...do you suggest a solution? I reverted your patch and then proceeded with > > > my work, so apologies for not having the dmesg log, if you need it I'll replicate, although > > > there is some subtest which has consistently been spitting out a warning on my Apple M3 Linux > > > VM and I have been just ignoring it for long : ) > > > > > This is because CONFIG_TEST_VMALLOC=y is built-in in your case. > > Actually, I have a usecase for test_vmalloc :) I wanted to test my > patches through the test, but the problem is that the moment it fails, > it gets unloaded and then I cannot use the test at runtime. Currently it > is set to =m in my .config. > Ah. It is unloaded itself when tests are done. Do not use the test_vmalloc.ko directly. Instead use test_vmalloc.sh script. It is located here: ./tools/testing/selftests/mm/test_vmalloc.sh Example: sudo ./test_vmalloc.sh nr_threads=HOW-MANY-CPU-YOU-have(digit) run_test_mask=7 -- Uladzislau Rezki ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution 2025-06-13 14:37 ` Uladzislau Rezki @ 2025-06-14 10:27 ` Dev Jain 2025-06-14 15:35 ` Uladzislau Rezki 0 siblings, 1 reply; 18+ messages in thread From: Dev Jain @ 2025-06-14 10:27 UTC (permalink / raw) To: Uladzislau Rezki Cc: Andrew Morton, linux-mm, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko, Ryan Roberts On 13/06/25 8:07 pm, Uladzislau Rezki wrote: > On Fri, Jun 13, 2025 at 07:53:37PM +0530, Dev Jain wrote: >> On 13/06/25 7:49 pm, Uladzislau Rezki wrote: >>> Hello, Dev! >>> >>>> On 17/04/25 9:42 pm, Uladzislau Rezki (Sony) wrote: >>>>> This patch removes the dependency on module loading ("m") >>>>> for the vmalloc test suite, enabling it to be built directly >>>>> into the kernel, so both ("=m") and ("=y") are supported. >>>>> >>>>> Motivation: >>>>> - Faster debugging/testing of vmalloc code; >>>>> - It allows to configure the test via kernel-boot parameters. >>>>> >>>>> Configuration example: >>>>> test_vmalloc.nr_threads=64 >>>>> test_vmalloc.run_test_mask=7 >>>>> test_vmalloc.sequential_test_order=1 >>>>> >>>>> Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> >>>>> --- >>>>> lib/Kconfig.debug | 3 +-- >>>>> lib/test_vmalloc.c | 5 +++-- >>>>> 2 files changed, 4 insertions(+), 4 deletions(-) >>>>> >>>>> diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug >>>>> index f9051ab610d54..166b9d830a85a 100644 >>>>> --- a/lib/Kconfig.debug >>>>> +++ b/lib/Kconfig.debug >>>>> @@ -2574,8 +2574,7 @@ config TEST_BITOPS >>>>> config TEST_VMALLOC >>>>> tristate "Test module for stress/performance analysis of vmalloc allocator" >>>>> default n >>>>> - depends on MMU >>>>> - depends on m >>>>> + depends on MMU >>>>> help >>>>> This builds the "test_vmalloc" module that should be used for >>>>> stress and performance analysis. So, any new change for vmalloc >>>>> diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c >>>>> index 4ab23e5e772d0..6d65ef725d42c 100644 >>>>> --- a/lib/test_vmalloc.c >>>>> +++ b/lib/test_vmalloc.c >>>>> @@ -591,10 +591,11 @@ static void do_concurrent_test(void) >>>>> kvfree(tdriver); >>>>> } >>>>> -static int vmalloc_test_init(void) >>>>> +static int __init vmalloc_test_init(void) >>>>> { >>>>> do_concurrent_test(); >>>>> - return -EAGAIN; /* Fail will directly unload the module */ >>>>> + /* Fail will directly unload the module */ >>>>> + return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN; >>>>> } >>>>> module_init(vmalloc_test_init) >>>> Hi Uladzislau, >>>> >>>> Post this patch I am seeing the following problem: while booting the kernel, this test is >>>> executed and a particular testcase triggers a warning splat from the kernel, then the boot >>>> gets stuck for 3-4 seconds, then proceeds. I am not an expert on modules so I don't know >>>> what happened but I am guessing that when the test fails, the delay is due to module >>>> unloading time...do you suggest a solution? I reverted your patch and then proceeded with >>>> my work, so apologies for not having the dmesg log, if you need it I'll replicate, although >>>> there is some subtest which has consistently been spitting out a warning on my Apple M3 Linux >>>> VM and I have been just ignoring it for long : ) >>>> >>> This is because CONFIG_TEST_VMALLOC=y is built-in in your case. >> Actually, I have a usecase for test_vmalloc :) I wanted to test my >> patches through the test, but the problem is that the moment it fails, >> it gets unloaded and then I cannot use the test at runtime. Currently it >> is set to =m in my .config. >> > Ah. It is unloaded itself when tests are done. Do not use the > test_vmalloc.ko directly. Instead use test_vmalloc.sh script. > > It is located here: ./tools/testing/selftests/mm/test_vmalloc.sh > > Example: sudo ./test_vmalloc.sh nr_threads=HOW-MANY-CPU-YOU-have(digit) run_test_mask=7 This is what I have been doing. Before your patch, I was able to run the test multiple times, even in case of failure. After the patch, it fails at boot time and unloads. > > -- > Uladzislau Rezki ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution 2025-06-14 10:27 ` Dev Jain @ 2025-06-14 15:35 ` Uladzislau Rezki 0 siblings, 0 replies; 18+ messages in thread From: Uladzislau Rezki @ 2025-06-14 15:35 UTC (permalink / raw) To: Dev Jain Cc: Uladzislau Rezki, Andrew Morton, linux-mm, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko, Ryan Roberts On Sat, Jun 14, 2025 at 03:57:26PM +0530, Dev Jain wrote: > > On 13/06/25 8:07 pm, Uladzislau Rezki wrote: > > On Fri, Jun 13, 2025 at 07:53:37PM +0530, Dev Jain wrote: > > > On 13/06/25 7:49 pm, Uladzislau Rezki wrote: > > > > Hello, Dev! > > > > > > > > > On 17/04/25 9:42 pm, Uladzislau Rezki (Sony) wrote: > > > > > > This patch removes the dependency on module loading ("m") > > > > > > for the vmalloc test suite, enabling it to be built directly > > > > > > into the kernel, so both ("=m") and ("=y") are supported. > > > > > > > > > > > > Motivation: > > > > > > - Faster debugging/testing of vmalloc code; > > > > > > - It allows to configure the test via kernel-boot parameters. > > > > > > > > > > > > Configuration example: > > > > > > test_vmalloc.nr_threads=64 > > > > > > test_vmalloc.run_test_mask=7 > > > > > > test_vmalloc.sequential_test_order=1 > > > > > > > > > > > > Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> > > > > > > --- > > > > > > lib/Kconfig.debug | 3 +-- > > > > > > lib/test_vmalloc.c | 5 +++-- > > > > > > 2 files changed, 4 insertions(+), 4 deletions(-) > > > > > > > > > > > > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug > > > > > > index f9051ab610d54..166b9d830a85a 100644 > > > > > > --- a/lib/Kconfig.debug > > > > > > +++ b/lib/Kconfig.debug > > > > > > @@ -2574,8 +2574,7 @@ config TEST_BITOPS > > > > > > config TEST_VMALLOC > > > > > > tristate "Test module for stress/performance analysis of vmalloc allocator" > > > > > > default n > > > > > > - depends on MMU > > > > > > - depends on m > > > > > > + depends on MMU > > > > > > help > > > > > > This builds the "test_vmalloc" module that should be used for > > > > > > stress and performance analysis. So, any new change for vmalloc > > > > > > diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c > > > > > > index 4ab23e5e772d0..6d65ef725d42c 100644 > > > > > > --- a/lib/test_vmalloc.c > > > > > > +++ b/lib/test_vmalloc.c > > > > > > @@ -591,10 +591,11 @@ static void do_concurrent_test(void) > > > > > > kvfree(tdriver); > > > > > > } > > > > > > -static int vmalloc_test_init(void) > > > > > > +static int __init vmalloc_test_init(void) > > > > > > { > > > > > > do_concurrent_test(); > > > > > > - return -EAGAIN; /* Fail will directly unload the module */ > > > > > > + /* Fail will directly unload the module */ > > > > > > + return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN; > > > > > > } > > > > > > module_init(vmalloc_test_init) > > > > > Hi Uladzislau, > > > > > > > > > > Post this patch I am seeing the following problem: while booting the kernel, this test is > > > > > executed and a particular testcase triggers a warning splat from the kernel, then the boot > > > > > gets stuck for 3-4 seconds, then proceeds. I am not an expert on modules so I don't know > > > > > what happened but I am guessing that when the test fails, the delay is due to module > > > > > unloading time...do you suggest a solution? I reverted your patch and then proceeded with > > > > > my work, so apologies for not having the dmesg log, if you need it I'll replicate, although > > > > > there is some subtest which has consistently been spitting out a warning on my Apple M3 Linux > > > > > VM and I have been just ignoring it for long : ) > > > > > > > > > This is because CONFIG_TEST_VMALLOC=y is built-in in your case. > > > Actually, I have a usecase for test_vmalloc :) I wanted to test my > > > patches through the test, but the problem is that the moment it fails, > > > it gets unloaded and then I cannot use the test at runtime. Currently it > > > is set to =m in my .config. > > > > > Ah. It is unloaded itself when tests are done. Do not use the > > test_vmalloc.ko directly. Instead use test_vmalloc.sh script. > > > > It is located here: ./tools/testing/selftests/mm/test_vmalloc.sh > > > > Example: sudo ./test_vmalloc.sh nr_threads=HOW-MANY-CPU-YOU-have(digit) run_test_mask=7 > > This is what I have been doing. Before your patch, I was able to run the test multiple > times, even in case of failure. After the patch, it fails at boot time and unloads. > If it is set as "m" it will not run during boot, because it will be compiled as a kernel module. Unless your system during a boot process loads all modules. But even then you can trigger the test how many times you want. It must be set as "y" in your system. Then yes, it executes only once during the boot. If you want to execute it many times set it as "m". Could you please do below: urezki@pc638:~/data/raid0/coding/linux.git$ cat .config | grep VMALLOC # CONFIG_DEBUG_PERF_USE_VMALLOC is not set CONFIG_HAVE_ARCH_HUGE_VMALLOC=y CONFIG_VIDEOBUF2_VMALLOC=m CONFIG_HAVE_ARCH_KASAN_VMALLOC=y CONFIG_TEST_VMALLOC=y urezki@pc638:~/data/raid0/coding/linux.git$ -- Uladzislau Rezki ^ permalink raw reply [flat|nested] 18+ messages in thread
* [PATCH 3/4] MAINTAINERS: Add test_vmalloc.c to VMALLOC section 2025-04-17 16:12 [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup Uladzislau Rezki (Sony) 2025-04-17 16:12 ` [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution Uladzislau Rezki (Sony) @ 2025-04-17 16:12 ` Uladzislau Rezki (Sony) 2025-04-18 15:01 ` Baoquan He 2025-04-17 16:12 ` [PATCH 4/4] vmalloc: Align nr_vmalloc_pages and vmap_lazy_nr Uladzislau Rezki (Sony) ` (2 subsequent siblings) 4 siblings, 1 reply; 18+ messages in thread From: Uladzislau Rezki (Sony) @ 2025-04-17 16:12 UTC (permalink / raw) To: Andrew Morton Cc: linux-mm, LKML, Baoquan He, Christoph Hellwig, Uladzislau Rezki, Oleksiy Avramchenko A vmalloc subsystem includes "lib/test_vmalloc.c" test suite. Add an "F:" entry under VMALLOC section to track this file as part of the subsystem. Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> --- MAINTAINERS | 1 + 1 file changed, 1 insertion(+) diff --git a/MAINTAINERS b/MAINTAINERS index 17ed0b5ffdd20..8e62b09e4c9e4 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -25874,6 +25874,7 @@ W: http://www.linux-mm.org T: git git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm F: include/linux/vmalloc.h F: mm/vmalloc.c +F: lib/test_vmalloc.c VME SUBSYSTEM L: linux-kernel@vger.kernel.org -- 2.39.5 ^ permalink raw reply related [flat|nested] 18+ messages in thread
* Re: [PATCH 3/4] MAINTAINERS: Add test_vmalloc.c to VMALLOC section 2025-04-17 16:12 ` [PATCH 3/4] MAINTAINERS: Add test_vmalloc.c to VMALLOC section Uladzislau Rezki (Sony) @ 2025-04-18 15:01 ` Baoquan He 0 siblings, 0 replies; 18+ messages in thread From: Baoquan He @ 2025-04-18 15:01 UTC (permalink / raw) To: Uladzislau Rezki (Sony) Cc: Andrew Morton, linux-mm, LKML, Christoph Hellwig, Oleksiy Avramchenko On 04/17/25 at 06:12pm, Uladzislau Rezki (Sony) wrote: > A vmalloc subsystem includes "lib/test_vmalloc.c" test suite. > Add an "F:" entry under VMALLOC section to track this file as > part of the subsystem. > > Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> > --- > MAINTAINERS | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/MAINTAINERS b/MAINTAINERS > index 17ed0b5ffdd20..8e62b09e4c9e4 100644 > --- a/MAINTAINERS > +++ b/MAINTAINERS > @@ -25874,6 +25874,7 @@ W: http://www.linux-mm.org > T: git git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm > F: include/linux/vmalloc.h > F: mm/vmalloc.c > +F: lib/test_vmalloc.c Reviewed-by: Baoquan He <bhe@redhat.com> > > VME SUBSYSTEM > L: linux-kernel@vger.kernel.org > -- > 2.39.5 > ^ permalink raw reply [flat|nested] 18+ messages in thread
* [PATCH 4/4] vmalloc: Align nr_vmalloc_pages and vmap_lazy_nr 2025-04-17 16:12 [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup Uladzislau Rezki (Sony) 2025-04-17 16:12 ` [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution Uladzislau Rezki (Sony) 2025-04-17 16:12 ` [PATCH 3/4] MAINTAINERS: Add test_vmalloc.c to VMALLOC section Uladzislau Rezki (Sony) @ 2025-04-17 16:12 ` Uladzislau Rezki (Sony) 2025-04-18 13:26 ` Baoquan He 2025-04-24 13:38 ` Adrian Huang12 2025-04-18 18:12 ` [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup kernel test robot 2025-04-24 13:35 ` Adrian Huang12 4 siblings, 2 replies; 18+ messages in thread From: Uladzislau Rezki (Sony) @ 2025-04-17 16:12 UTC (permalink / raw) To: Andrew Morton Cc: linux-mm, LKML, Baoquan He, Christoph Hellwig, Uladzislau Rezki, Oleksiy Avramchenko, Mateusz Guzik Currently both atomics share one cache-line: <snip> ... ffffffff83eab400 b vmap_lazy_nr ffffffff83eab408 b nr_vmalloc_pages ... <snip> those are global variables and they are only 8 bytes apart. Since they are modified by different threads this causes a false sharing. This can lead to a performance drop due to unnecessary cache invalidations. After this patch it is aligned to a cache line boundary: <snip> ... ffffffff8260a600 d vmap_lazy_nr ffffffff8260a640 d nr_vmalloc_pages ... <snip> Cc: Mateusz Guzik <mjguzik@gmail.com> Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> --- mm/vmalloc.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/mm/vmalloc.c b/mm/vmalloc.c index 77da4613f07ff..54f60d62051da 100644 --- a/mm/vmalloc.c +++ b/mm/vmalloc.c @@ -1008,7 +1008,8 @@ static BLOCKING_NOTIFIER_HEAD(vmap_notify_list); static void drain_vmap_area_work(struct work_struct *work); static DECLARE_WORK(drain_vmap_work, drain_vmap_area_work); -static atomic_long_t nr_vmalloc_pages; +static __cacheline_aligned_in_smp atomic_long_t nr_vmalloc_pages; +static __cacheline_aligned_in_smp atomic_long_t vmap_lazy_nr; unsigned long vmalloc_nr_pages(void) { @@ -2117,8 +2118,6 @@ static unsigned long lazy_max_pages(void) return log * (32UL * 1024 * 1024 / PAGE_SIZE); } -static atomic_long_t vmap_lazy_nr = ATOMIC_LONG_INIT(0); - /* * Serialize vmap purging. There is no actual critical section protected * by this lock, but we want to avoid concurrent calls for performance -- 2.39.5 ^ permalink raw reply related [flat|nested] 18+ messages in thread
* Re: [PATCH 4/4] vmalloc: Align nr_vmalloc_pages and vmap_lazy_nr 2025-04-17 16:12 ` [PATCH 4/4] vmalloc: Align nr_vmalloc_pages and vmap_lazy_nr Uladzislau Rezki (Sony) @ 2025-04-18 13:26 ` Baoquan He 2025-04-24 13:38 ` Adrian Huang12 1 sibling, 0 replies; 18+ messages in thread From: Baoquan He @ 2025-04-18 13:26 UTC (permalink / raw) To: Uladzislau Rezki (Sony) Cc: Andrew Morton, linux-mm, LKML, Christoph Hellwig, Oleksiy Avramchenko, Mateusz Guzik On 04/17/25 at 06:12pm, Uladzislau Rezki (Sony) wrote: > Currently both atomics share one cache-line: > > <snip> > ... > ffffffff83eab400 b vmap_lazy_nr > ffffffff83eab408 b nr_vmalloc_pages > ... > <snip> > > those are global variables and they are only 8 bytes apart. > Since they are modified by different threads this causes a > false sharing. This can lead to a performance drop due to > unnecessary cache invalidations. > > After this patch it is aligned to a cache line boundary: > > <snip> > ... > ffffffff8260a600 d vmap_lazy_nr > ffffffff8260a640 d nr_vmalloc_pages A great catch and improvement. Reviewed-by: Baoquan He <bhe@redhat.com> > ... > <snip> > > Cc: Mateusz Guzik <mjguzik@gmail.com> > Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> > --- > mm/vmalloc.c | 5 ++--- > 1 file changed, 2 insertions(+), 3 deletions(-) > > diff --git a/mm/vmalloc.c b/mm/vmalloc.c > index 77da4613f07ff..54f60d62051da 100644 > --- a/mm/vmalloc.c > +++ b/mm/vmalloc.c > @@ -1008,7 +1008,8 @@ static BLOCKING_NOTIFIER_HEAD(vmap_notify_list); > static void drain_vmap_area_work(struct work_struct *work); > static DECLARE_WORK(drain_vmap_work, drain_vmap_area_work); > > -static atomic_long_t nr_vmalloc_pages; > +static __cacheline_aligned_in_smp atomic_long_t nr_vmalloc_pages; > +static __cacheline_aligned_in_smp atomic_long_t vmap_lazy_nr; > > unsigned long vmalloc_nr_pages(void) > { > @@ -2117,8 +2118,6 @@ static unsigned long lazy_max_pages(void) > return log * (32UL * 1024 * 1024 / PAGE_SIZE); > } > > -static atomic_long_t vmap_lazy_nr = ATOMIC_LONG_INIT(0); > - > /* > * Serialize vmap purging. There is no actual critical section protected > * by this lock, but we want to avoid concurrent calls for performance > -- > 2.39.5 > ^ permalink raw reply [flat|nested] 18+ messages in thread
* RE: [PATCH 4/4] vmalloc: Align nr_vmalloc_pages and vmap_lazy_nr 2025-04-17 16:12 ` [PATCH 4/4] vmalloc: Align nr_vmalloc_pages and vmap_lazy_nr Uladzislau Rezki (Sony) 2025-04-18 13:26 ` Baoquan He @ 2025-04-24 13:38 ` Adrian Huang12 1 sibling, 0 replies; 18+ messages in thread From: Adrian Huang12 @ 2025-04-24 13:38 UTC (permalink / raw) To: Uladzislau Rezki (Sony), Andrew Morton Cc: linux-mm@kvack.org, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko, Mateusz Guzik > -----Original Message----- > From: owner-linux-mm@kvack.org <owner-linux-mm@kvack.org> On Behalf > Of Uladzislau Rezki (Sony) > Sent: Friday, April 18, 2025 12:12 AM > To: Andrew Morton <akpm@linux-foundation.org> > Cc: linux-mm@kvack.org; LKML <linux-kernel@vger.kernel.org>; Baoquan He > <bhe@redhat.com>; Christoph Hellwig <hch@infradead.org>; Uladzislau Rezki > <urezki@gmail.com>; Oleksiy Avramchenko <oleksiy.avramchenko@sony.com>; > Mateusz Guzik <mjguzik@gmail.com> > Subject: [External] [PATCH 4/4] vmalloc: Align nr_vmalloc_pages and > vmap_lazy_nr > > Currently both atomics share one cache-line: > > <snip> > ... > ffffffff83eab400 b vmap_lazy_nr > ffffffff83eab408 b nr_vmalloc_pages > ... > <snip> > > those are global variables and they are only 8 bytes apart. > Since they are modified by different threads this causes a false sharing. This > can lead to a performance drop due to unnecessary cache invalidations. > > After this patch it is aligned to a cache line boundary: > > <snip> > ... > ffffffff8260a600 d vmap_lazy_nr > ffffffff8260a640 d nr_vmalloc_pages > ... > <snip> > > Cc: Mateusz Guzik <mjguzik@gmail.com> > Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com> > --- > mm/vmalloc.c | 5 ++--- > 1 file changed, 2 insertions(+), 3 deletions(-) Reviewed-by: Adrian Huang <ahuang12@lenovo.com> Tested-by: Adrian Huang <ahuang12@lenovo.com> > diff --git a/mm/vmalloc.c b/mm/vmalloc.c index > 77da4613f07ff..54f60d62051da 100644 > --- a/mm/vmalloc.c > +++ b/mm/vmalloc.c > @@ -1008,7 +1008,8 @@ static BLOCKING_NOTIFIER_HEAD(vmap_notify_list); > static void drain_vmap_area_work(struct work_struct *work); static > DECLARE_WORK(drain_vmap_work, drain_vmap_area_work); > > -static atomic_long_t nr_vmalloc_pages; > +static __cacheline_aligned_in_smp atomic_long_t nr_vmalloc_pages; > +static __cacheline_aligned_in_smp atomic_long_t vmap_lazy_nr; > > unsigned long vmalloc_nr_pages(void) > { > @@ -2117,8 +2118,6 @@ static unsigned long lazy_max_pages(void) > return log * (32UL * 1024 * 1024 / PAGE_SIZE); } > > -static atomic_long_t vmap_lazy_nr = ATOMIC_LONG_INIT(0); > - > /* > * Serialize vmap purging. There is no actual critical section protected > * by this lock, but we want to avoid concurrent calls for performance > -- > 2.39.5 > ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup 2025-04-17 16:12 [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup Uladzislau Rezki (Sony) ` (2 preceding siblings ...) 2025-04-17 16:12 ` [PATCH 4/4] vmalloc: Align nr_vmalloc_pages and vmap_lazy_nr Uladzislau Rezki (Sony) @ 2025-04-18 18:12 ` kernel test robot 2025-04-22 12:59 ` Uladzislau Rezki 2025-04-24 13:35 ` Adrian Huang12 4 siblings, 1 reply; 18+ messages in thread From: kernel test robot @ 2025-04-18 18:12 UTC (permalink / raw) To: Uladzislau Rezki (Sony), Andrew Morton Cc: oe-kbuild-all, Linux Memory Management List, LKML, Baoquan He, Christoph Hellwig, Uladzislau Rezki, Oleksiy Avramchenko Hi Uladzislau, kernel test robot noticed the following build errors: [auto build test ERROR on akpm-mm/mm-nonmm-unstable] [also build test ERROR on linus/master hch-configfs/for-next v6.15-rc2 next-20250417] [cannot apply to akpm-mm/mm-everything] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in https://git-scm.com/docs/git-format-patch#_base_tree_information] url: https://github.com/intel-lab-lkp/linux/commits/Uladzislau-Rezki-Sony/lib-test_vmalloc-c-Allow-built-in-execution/20250418-001446 base: https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-nonmm-unstable patch link: https://lore.kernel.org/r/20250417161216.88318-1-urezki%40gmail.com patch subject: [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup config: sparc64-randconfig-001-20250419 (https://download.01.org/0day-ci/archive/20250419/202504190146.gTBIrktk-lkp@intel.com/config) compiler: sparc64-linux-gcc (GCC) 8.5.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250419/202504190146.gTBIrktk-lkp@intel.com/reproduce) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp@intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202504190146.gTBIrktk-lkp@intel.com/ All error/warnings (new ones prefixed by >>): In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:69:8: error: redefinition of 'struct srcu_usage' struct srcu_usage { ^~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:52:8: note: originally defined here struct srcu_usage { }; ^~~~~~~~~~ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:102:8: error: redefinition of 'struct srcu_struct' struct srcu_struct { ^~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:16:8: note: originally defined here struct srcu_struct { ^~~~~~~~~~~ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:149: warning: "__SRCU_USAGE_INIT" redefined #define __SRCU_USAGE_INIT(name) \ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:53: note: this is the location of the previous definition #define __SRCU_USAGE_INIT(name) { } In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:167: warning: "__SRCU_STRUCT_INIT" redefined #define __SRCU_STRUCT_INIT(name, usage_name, pcpu_name) \ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:34: note: this is the location of the previous definition #define __SRCU_STRUCT_INIT(name, __ignored, ___ignored) \ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:207: warning: "DEFINE_SRCU" redefined #define DEFINE_SRCU(name) __DEFINE_SRCU(name, /* not static */) In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:46: note: this is the location of the previous definition #define DEFINE_SRCU(name) \ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:208: warning: "DEFINE_STATIC_SRCU" redefined #define DEFINE_STATIC_SRCU(name) __DEFINE_SRCU(name, static) In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:48: note: this is the location of the previous definition #define DEFINE_STATIC_SRCU(name) \ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:210:5: error: conflicting types for '__srcu_read_lock' int __srcu_read_lock(struct srcu_struct *ssp) __acquires(ssp); ^~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:63:19: note: previous definition of '__srcu_read_lock' was here static inline int __srcu_read_lock(struct srcu_struct *ssp) ^~~~~~~~~~~~~~~~ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:211:6: error: conflicting types for 'synchronize_srcu_expedited' void synchronize_srcu_expedited(struct srcu_struct *ssp); ^~~~~~~~~~~~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:99:20: note: previous definition of 'synchronize_srcu_expedited' was here static inline void synchronize_srcu_expedited(struct srcu_struct *ssp) ^~~~~~~~~~~~~~~~~~~~~~~~~~ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:212:6: error: conflicting types for 'srcu_barrier' void srcu_barrier(struct srcu_struct *ssp); ^~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:104:20: note: previous definition of 'srcu_barrier' was here static inline void srcu_barrier(struct srcu_struct *ssp) ^~~~~~~~~~~~ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:213:6: error: conflicting types for 'srcu_torture_stats_print' void srcu_torture_stats_print(struct srcu_struct *ssp, char *tt, char *tf); ^~~~~~~~~~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:113:20: note: previous definition of 'srcu_torture_stats_print' was here static inline void srcu_torture_stats_print(struct srcu_struct *ssp, ^~~~~~~~~~~~~~~~~~~~~~~~ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:217:20: error: conflicting types for '__srcu_ptr_to_ctr' static inline bool __srcu_ptr_to_ctr(struct srcu_struct *ssp, struct srcu_ctr __percpu *scpp) ^~~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:76:20: note: previous definition of '__srcu_ptr_to_ctr' was here static inline bool __srcu_ptr_to_ctr(struct srcu_struct *ssp, struct srcu_ctr __percpu *scpp) ^~~~~~~~~~~~~~~~~ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:224:41: error: conflicting types for '__srcu_ctr_to_ptr' static inline struct srcu_ctr __percpu *__srcu_ctr_to_ptr(struct srcu_struct *ssp, int idx) ^~~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:81:41: note: previous definition of '__srcu_ctr_to_ptr' was here static inline struct srcu_ctr __percpu *__srcu_ctr_to_ptr(struct srcu_struct *ssp, int idx) ^~~~~~~~~~~~~~~~~ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:243:41: error: conflicting types for '__srcu_read_lock_fast' static inline struct srcu_ctr __percpu *__srcu_read_lock_fast(struct srcu_struct *ssp) ^~~~~~~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:86:41: note: previous definition of '__srcu_read_lock_fast' was here static inline struct srcu_ctr __percpu *__srcu_read_lock_fast(struct srcu_struct *ssp) ^~~~~~~~~~~~~~~~~~~~~ In file included from lib/test_vmalloc.c:18: >> include/linux/srcutree.h:271:20: error: conflicting types for '__srcu_read_unlock_fast' static inline void __srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_ctr __percpu *scp) ^~~~~~~~~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:91:20: note: previous definition of '__srcu_read_unlock_fast' was here static inline void __srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_ctr __percpu *scp) ^~~~~~~~~~~~~~~~~~~~~~~ >> include/linux/srcutiny.h:96:31: error: static declaration of '__srcu_read_lock' follows non-static declaration #define __srcu_read_lock_lite __srcu_read_lock ^~~~~~~~~~~~~~~~ include/linux/srcutree.h:291:19: note: in expansion of macro '__srcu_read_lock_lite' static inline int __srcu_read_lock_lite(struct srcu_struct *ssp) ^~~~~~~~~~~~~~~~~~~~~ In file included from lib/test_vmalloc.c:18: include/linux/srcutree.h:210:5: note: previous declaration of '__srcu_read_lock' was here int __srcu_read_lock(struct srcu_struct *ssp) __acquires(ssp); ^~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: >> include/linux/srcutiny.h:97:33: error: conflicting types for '__srcu_read_unlock' #define __srcu_read_unlock_lite __srcu_read_unlock ^~~~~~~~~~~~~~~~~~ include/linux/srcutree.h:312:20: note: in expansion of macro '__srcu_read_unlock_lite' static inline void __srcu_read_unlock_lite(struct srcu_struct *ssp, int idx) ^~~~~~~~~~~~~~~~~~~~~~~ In file included from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcu.h:56:6: note: previous declaration of '__srcu_read_unlock' was here void __srcu_read_unlock(struct srcu_struct *ssp, int idx) __releases(ssp); ^~~~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: >> include/linux/srcutiny.h:110:56: error: expected identifier or '(' before 'do' #define srcu_check_read_flavor_force(ssp, read_flavor) do { } while (0) ^~ include/linux/srcutree.h:324:20: note: in expansion of macro 'srcu_check_read_flavor_force' static inline void srcu_check_read_flavor_force(struct srcu_struct *ssp, int read_flavor) ^~~~~~~~~~~~~~~~~~~~~~~~~~~~ >> include/linux/srcutiny.h:110:63: error: expected identifier or '(' before 'while' #define srcu_check_read_flavor_force(ssp, read_flavor) do { } while (0) ^~~~~ include/linux/srcutree.h:324:20: note: in expansion of macro 'srcu_check_read_flavor_force' static inline void srcu_check_read_flavor_force(struct srcu_struct *ssp, int read_flavor) ^~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/linux/srcutiny.h:109:50: error: expected identifier or '(' before 'do' #define srcu_check_read_flavor(ssp, read_flavor) do { } while (0) ^~ include/linux/srcutree.h:336:20: note: in expansion of macro 'srcu_check_read_flavor' static inline void srcu_check_read_flavor(struct srcu_struct *ssp, int read_flavor) ^~~~~~~~~~~~~~~~~~~~~~ include/linux/srcutiny.h:109:57: error: expected identifier or '(' before 'while' #define srcu_check_read_flavor(ssp, read_flavor) do { } while (0) ^~~~~ include/linux/srcutree.h:336:20: note: in expansion of macro 'srcu_check_read_flavor' static inline void srcu_check_read_flavor(struct srcu_struct *ssp, int read_flavor) ^~~~~~~~~~~~~~~~~~~~~~ lib/test_vmalloc.c: In function 'test_func': >> lib/test_vmalloc.c:460:19: error: passing argument 1 of 'synchronize_srcu' from incompatible pointer type [-Werror=incompatible-pointer-types] synchronize_srcu(&prepare_for_test_srcu); ^~~~~~~~~~~~~~~~~~~~~~ In file included from include/linux/srcu.h:59, from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcutiny.h:55:43: note: expected 'struct srcu_struct *' but argument is of type 'struct srcu_struct *' void synchronize_srcu(struct srcu_struct *ssp); ~~~~~~~~~~~~~~~~~~~~^~~ lib/test_vmalloc.c: In function 'do_concurrent_test': >> lib/test_vmalloc.c:538:23: error: passing argument 1 of 'srcu_read_lock' from incompatible pointer type [-Werror=incompatible-pointer-types] idx = srcu_read_lock(&prepare_for_test_srcu); ^~~~~~~~~~~~~~~~~~~~~~ In file included from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcu.h:247:54: note: expected 'struct srcu_struct *' but argument is of type 'struct srcu_struct *' static inline int srcu_read_lock(struct srcu_struct *ssp) __acquires(ssp) ~~~~~~~~~~~~~~~~~~~~^~~ lib/test_vmalloc.c:555:19: error: passing argument 1 of 'srcu_read_unlock' from incompatible pointer type [-Werror=incompatible-pointer-types] srcu_read_unlock(&prepare_for_test_srcu, idx); ^~~~~~~~~~~~~~~~~~~~~~ In file included from include/linux/notifier.h:16, from include/linux/memory_hotplug.h:7, from include/linux/mmzone.h:1494, from include/linux/gfp.h:7, from include/linux/umh.h:4, from include/linux/kmod.h:9, from include/linux/module.h:17, from lib/test_vmalloc.c:9: include/linux/srcu.h:397:57: note: expected 'struct srcu_struct *' but argument is of type 'struct srcu_struct *' static inline void srcu_read_unlock(struct srcu_struct *ssp, int idx) ~~~~~~~~~~~~~~~~~~~~^~~ cc1: some warnings being treated as errors .. vim +69 include/linux/srcutree.h da915ad5cf25b5 Paul E. McKenney 2017-04-05 65 da915ad5cf25b5 Paul E. McKenney 2017-04-05 66 /* 95433f7263011e Paul E. McKenney 2023-03-16 67 * Per-SRCU-domain structure, update-side data linked from srcu_struct. da915ad5cf25b5 Paul E. McKenney 2017-04-05 68 */ 95433f7263011e Paul E. McKenney 2023-03-16 @69 struct srcu_usage { 2ec303113d9789 Paul E. McKenney 2022-01-21 70 struct srcu_node *node; /* Combining tree. */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 71 struct srcu_node *level[RCU_NUM_LVLS + 1]; da915ad5cf25b5 Paul E. McKenney 2017-04-05 72 /* First node at each level. */ 994f706872e6ce Paul E. McKenney 2022-01-24 73 int srcu_size_state; /* Small-to-big transition state. */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 74 struct mutex srcu_cb_mutex; /* Serialize CB preparation. */ 994f706872e6ce Paul E. McKenney 2022-01-24 75 spinlock_t __private lock; /* Protect counters and size state. */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 76 struct mutex srcu_gp_mutex; /* Serialize GP work. */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 77 unsigned long srcu_gp_seq; /* Grace-period seq #. */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 78 unsigned long srcu_gp_seq_needed; /* Latest gp_seq needed. */ 1e9a038b7fe9a8 Paul E. McKenney 2017-04-24 79 unsigned long srcu_gp_seq_needed_exp; /* Furthest future exp GP. */ 282d8998e9979c Paul E. McKenney 2022-03-08 80 unsigned long srcu_gp_start; /* Last GP start timestamp (jiffies) */ 22607d66bbc3e8 Paul E. McKenney 2017-04-25 81 unsigned long srcu_last_gp_end; /* Last GP end timestamp (ns) */ 9f2e91d94c9155 Paul E. McKenney 2022-01-27 82 unsigned long srcu_size_jiffies; /* Current contention-measurement interval. */ 9f2e91d94c9155 Paul E. McKenney 2022-01-27 83 unsigned long srcu_n_lock_retries; /* Contention events in current interval. */ 282d8998e9979c Paul E. McKenney 2022-03-08 84 unsigned long srcu_n_exp_nodelay; /* # expedited no-delays in current GP phase. */ 46470cf85d2b61 Paul E. McKenney 2022-01-27 85 bool sda_is_static; /* May ->sda be passed to free_percpu()? */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 86 unsigned long srcu_barrier_seq; /* srcu_barrier seq #. */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 87 struct mutex srcu_barrier_mutex; /* Serialize barrier ops. */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 88 struct completion srcu_barrier_completion; da915ad5cf25b5 Paul E. McKenney 2017-04-05 89 /* Awaken barrier rq at end. */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 90 atomic_t srcu_barrier_cpu_cnt; /* # CPUs not yet posting a */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 91 /* callback for the barrier */ da915ad5cf25b5 Paul E. McKenney 2017-04-05 92 /* operation. */ 282d8998e9979c Paul E. McKenney 2022-03-08 93 unsigned long reschedule_jiffies; 282d8998e9979c Paul E. McKenney 2022-03-08 94 unsigned long reschedule_count; d8be81735aa894 Paul E. McKenney 2017-03-25 95 struct delayed_work work; fd1b3f8e097b7f Paul E. McKenney 2023-03-17 96 struct srcu_struct *srcu_ssp; 95433f7263011e Paul E. McKenney 2023-03-16 97 }; 95433f7263011e Paul E. McKenney 2023-03-16 98 da915ad5cf25b5 Paul E. McKenney 2017-04-05 99 /* da915ad5cf25b5 Paul E. McKenney 2017-04-05 100 * Per-SRCU-domain structure, similar in function to rcu_state. da915ad5cf25b5 Paul E. McKenney 2017-04-05 101 */ d8be81735aa894 Paul E. McKenney 2017-03-25 @102 struct srcu_struct { 795e7efec6ea7e Paul E. McKenney 2024-12-19 103 struct srcu_ctr __percpu *srcu_ctrp; da915ad5cf25b5 Paul E. McKenney 2017-04-05 104 struct srcu_data __percpu *sda; /* Per-CPU srcu_data array. */ d8be81735aa894 Paul E. McKenney 2017-03-25 105 struct lockdep_map dep_map; 95433f7263011e Paul E. McKenney 2023-03-16 106 struct srcu_usage *srcu_sup; /* Update-side data. */ d8be81735aa894 Paul E. McKenney 2017-03-25 107 }; d8be81735aa894 Paul E. McKenney 2017-03-25 108 e15a19306004b3 Pingfan Liu 2023-01-04 109 // Values for size state variable (->srcu_size_state). Once the state e15a19306004b3 Pingfan Liu 2023-01-04 110 // has been set to SRCU_SIZE_ALLOC, the grace-period code advances through e15a19306004b3 Pingfan Liu 2023-01-04 111 // this state machine one step per grace period until the SRCU_SIZE_BIG state e15a19306004b3 Pingfan Liu 2023-01-04 112 // is reached. Otherwise, the state machine remains in the SRCU_SIZE_SMALL e15a19306004b3 Pingfan Liu 2023-01-04 113 // state indefinitely. e15a19306004b3 Pingfan Liu 2023-01-04 114 #define SRCU_SIZE_SMALL 0 // No srcu_node combining tree, ->node == NULL e15a19306004b3 Pingfan Liu 2023-01-04 115 #define SRCU_SIZE_ALLOC 1 // An srcu_node tree is being allocated, initialized, e15a19306004b3 Pingfan Liu 2023-01-04 116 // and then referenced by ->node. It will not be used. e15a19306004b3 Pingfan Liu 2023-01-04 117 #define SRCU_SIZE_WAIT_BARRIER 2 // The srcu_node tree starts being used by everything e15a19306004b3 Pingfan Liu 2023-01-04 118 // except call_srcu(), especially by srcu_barrier(). e15a19306004b3 Pingfan Liu 2023-01-04 119 // By the end of this state, all CPUs and threads e15a19306004b3 Pingfan Liu 2023-01-04 120 // are aware of this tree's existence. e15a19306004b3 Pingfan Liu 2023-01-04 121 #define SRCU_SIZE_WAIT_CALL 3 // The srcu_node tree starts being used by call_srcu(). e15a19306004b3 Pingfan Liu 2023-01-04 122 // By the end of this state, all of the call_srcu() e15a19306004b3 Pingfan Liu 2023-01-04 123 // invocations that were running on a non-boot CPU e15a19306004b3 Pingfan Liu 2023-01-04 124 // and using the boot CPU's callback queue will have e15a19306004b3 Pingfan Liu 2023-01-04 125 // completed. e15a19306004b3 Pingfan Liu 2023-01-04 126 #define SRCU_SIZE_WAIT_CBS1 4 // Don't trust the ->srcu_have_cbs[] grace-period e15a19306004b3 Pingfan Liu 2023-01-04 127 #define SRCU_SIZE_WAIT_CBS2 5 // sequence elements or the ->srcu_data_have_cbs[] e15a19306004b3 Pingfan Liu 2023-01-04 128 #define SRCU_SIZE_WAIT_CBS3 6 // CPU-bitmask elements until all four elements of e15a19306004b3 Pingfan Liu 2023-01-04 129 #define SRCU_SIZE_WAIT_CBS4 7 // each array have been initialized. e15a19306004b3 Pingfan Liu 2023-01-04 130 #define SRCU_SIZE_BIG 8 // The srcu_node combining tree is fully initialized e15a19306004b3 Pingfan Liu 2023-01-04 131 // and all aspects of it are being put to use. 994f706872e6ce Paul E. McKenney 2022-01-24 132 da915ad5cf25b5 Paul E. McKenney 2017-04-05 133 /* Values for state variable (bottom bits of ->srcu_gp_seq). */ d8be81735aa894 Paul E. McKenney 2017-03-25 134 #define SRCU_STATE_IDLE 0 d8be81735aa894 Paul E. McKenney 2017-03-25 135 #define SRCU_STATE_SCAN1 1 d8be81735aa894 Paul E. McKenney 2017-03-25 136 #define SRCU_STATE_SCAN2 2 d8be81735aa894 Paul E. McKenney 2017-03-25 137 29bc83e4d90546 JP Kobryn 2024-07-15 138 /* 29bc83e4d90546 JP Kobryn 2024-07-15 139 * Values for initializing gp sequence fields. Higher values allow wrap arounds to 29bc83e4d90546 JP Kobryn 2024-07-15 140 * occur earlier. 29bc83e4d90546 JP Kobryn 2024-07-15 141 * The second value with state is useful in the case of static initialization of 29bc83e4d90546 JP Kobryn 2024-07-15 142 * srcu_usage where srcu_gp_seq_needed is expected to have some state value in its 29bc83e4d90546 JP Kobryn 2024-07-15 143 * lower bits (or else it will appear to be already initialized within 29bc83e4d90546 JP Kobryn 2024-07-15 144 * the call check_init_srcu_struct()). 29bc83e4d90546 JP Kobryn 2024-07-15 145 */ 29bc83e4d90546 JP Kobryn 2024-07-15 146 #define SRCU_GP_SEQ_INITIAL_VAL ((0UL - 100UL) << RCU_SEQ_CTR_SHIFT) 29bc83e4d90546 JP Kobryn 2024-07-15 147 #define SRCU_GP_SEQ_INITIAL_VAL_WITH_STATE (SRCU_GP_SEQ_INITIAL_VAL - 1) 29bc83e4d90546 JP Kobryn 2024-07-15 148 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 @149 #define __SRCU_USAGE_INIT(name) \ d8be81735aa894 Paul E. McKenney 2017-03-25 150 { \ d633198088bd9e Paul E. McKenney 2017-10-10 151 .lock = __SPIN_LOCK_UNLOCKED(name.lock), \ 29bc83e4d90546 JP Kobryn 2024-07-15 152 .srcu_gp_seq = SRCU_GP_SEQ_INITIAL_VAL, \ 29bc83e4d90546 JP Kobryn 2024-07-15 153 .srcu_gp_seq_needed = SRCU_GP_SEQ_INITIAL_VAL_WITH_STATE, \ 29bc83e4d90546 JP Kobryn 2024-07-15 154 .srcu_gp_seq_needed_exp = SRCU_GP_SEQ_INITIAL_VAL, \ 4e6ea4ef56f942 Paul E. McKenney 2018-08-14 155 .work = __DELAYED_WORK_INITIALIZER(name.work, NULL, 0), \ 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 156 } 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 157 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 158 #define __SRCU_STRUCT_INIT_COMMON(name, usage_name) \ 95433f7263011e Paul E. McKenney 2023-03-16 159 .srcu_sup = &usage_name, \ f4d01a259374ef Paul E. McKenney 2023-03-17 160 __SRCU_DEP_MAP_INIT(name) f4d01a259374ef Paul E. McKenney 2023-03-17 161 95433f7263011e Paul E. McKenney 2023-03-16 162 #define __SRCU_STRUCT_INIT_MODULE(name, usage_name) \ f4d01a259374ef Paul E. McKenney 2023-03-17 163 { \ 95433f7263011e Paul E. McKenney 2023-03-16 164 __SRCU_STRUCT_INIT_COMMON(name, usage_name) \ d8be81735aa894 Paul E. McKenney 2017-03-25 165 } d8be81735aa894 Paul E. McKenney 2017-03-25 166 95433f7263011e Paul E. McKenney 2023-03-16 @167 #define __SRCU_STRUCT_INIT(name, usage_name, pcpu_name) \ f4d01a259374ef Paul E. McKenney 2023-03-17 168 { \ f4d01a259374ef Paul E. McKenney 2023-03-17 169 .sda = &pcpu_name, \ 795e7efec6ea7e Paul E. McKenney 2024-12-19 170 .srcu_ctrp = &pcpu_name.srcu_ctrs[0], \ 95433f7263011e Paul E. McKenney 2023-03-16 171 __SRCU_STRUCT_INIT_COMMON(name, usage_name) \ d8be81735aa894 Paul E. McKenney 2017-03-25 172 } d8be81735aa894 Paul E. McKenney 2017-03-25 173 d8be81735aa894 Paul E. McKenney 2017-03-25 174 /* d8be81735aa894 Paul E. McKenney 2017-03-25 175 * Define and initialize a srcu struct at build time. d8be81735aa894 Paul E. McKenney 2017-03-25 176 * Do -not- call init_srcu_struct() nor cleanup_srcu_struct() on it. d8be81735aa894 Paul E. McKenney 2017-03-25 177 * d8be81735aa894 Paul E. McKenney 2017-03-25 178 * Note that although DEFINE_STATIC_SRCU() hides the name from other d8be81735aa894 Paul E. McKenney 2017-03-25 179 * files, the per-CPU variable rules nevertheless require that the d8be81735aa894 Paul E. McKenney 2017-03-25 180 * chosen name be globally unique. These rules also prohibit use of d8be81735aa894 Paul E. McKenney 2017-03-25 181 * DEFINE_STATIC_SRCU() within a function. If these rules are too d8be81735aa894 Paul E. McKenney 2017-03-25 182 * restrictive, declare the srcu_struct manually. For example, in d8be81735aa894 Paul E. McKenney 2017-03-25 183 * each file: d8be81735aa894 Paul E. McKenney 2017-03-25 184 * d8be81735aa894 Paul E. McKenney 2017-03-25 185 * static struct srcu_struct my_srcu; d8be81735aa894 Paul E. McKenney 2017-03-25 186 * d8be81735aa894 Paul E. McKenney 2017-03-25 187 * Then, before the first use of each my_srcu, manually initialize it: d8be81735aa894 Paul E. McKenney 2017-03-25 188 * d8be81735aa894 Paul E. McKenney 2017-03-25 189 * init_srcu_struct(&my_srcu); d8be81735aa894 Paul E. McKenney 2017-03-25 190 * d8be81735aa894 Paul E. McKenney 2017-03-25 191 * See include/linux/percpu-defs.h for the rules on per-CPU variables. d8be81735aa894 Paul E. McKenney 2017-03-25 192 */ fe15b50cdeeebd Paul E. McKenney 2019-04-05 193 #ifdef MODULE fe15b50cdeeebd Paul E. McKenney 2019-04-05 194 # define __DEFINE_SRCU(name, is_static) \ 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 195 static struct srcu_usage name##_srcu_usage = __SRCU_USAGE_INIT(name##_srcu_usage); \ 95433f7263011e Paul E. McKenney 2023-03-16 196 is_static struct srcu_struct name = __SRCU_STRUCT_INIT_MODULE(name, name##_srcu_usage); \ db8f1471c61336 Alexander Aring 2022-01-26 197 extern struct srcu_struct * const __srcu_struct_##name; \ 056b89e7e69974 Joel Fernandes (Google 2019-04-11 198) struct srcu_struct * const __srcu_struct_##name \ fe15b50cdeeebd Paul E. McKenney 2019-04-05 199 __section("___srcu_struct_ptrs") = &name fe15b50cdeeebd Paul E. McKenney 2019-04-05 200 #else d8be81735aa894 Paul E. McKenney 2017-03-25 201 # define __DEFINE_SRCU(name, is_static) \ da915ad5cf25b5 Paul E. McKenney 2017-04-05 202 static DEFINE_PER_CPU(struct srcu_data, name##_srcu_data); \ 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 203 static struct srcu_usage name##_srcu_usage = __SRCU_USAGE_INIT(name##_srcu_usage); \ fe15b50cdeeebd Paul E. McKenney 2019-04-05 204 is_static struct srcu_struct name = \ 95433f7263011e Paul E. McKenney 2023-03-16 205 __SRCU_STRUCT_INIT(name, name##_srcu_usage, name##_srcu_data) fe15b50cdeeebd Paul E. McKenney 2019-04-05 206 #endif d8be81735aa894 Paul E. McKenney 2017-03-25 @207 #define DEFINE_SRCU(name) __DEFINE_SRCU(name, /* not static */) d8be81735aa894 Paul E. McKenney 2017-03-25 @208 #define DEFINE_STATIC_SRCU(name) __DEFINE_SRCU(name, static) d8be81735aa894 Paul E. McKenney 2017-03-25 209 443971156cebfc Paul E. McKenney 2025-01-13 @210 int __srcu_read_lock(struct srcu_struct *ssp) __acquires(ssp); aacb5d91ab1bfb Paul E. McKenney 2018-10-28 @211 void synchronize_srcu_expedited(struct srcu_struct *ssp); aacb5d91ab1bfb Paul E. McKenney 2018-10-28 @212 void srcu_barrier(struct srcu_struct *ssp); aacb5d91ab1bfb Paul E. McKenney 2018-10-28 @213 void srcu_torture_stats_print(struct srcu_struct *ssp, char *tt, char *tf); d8be81735aa894 Paul E. McKenney 2017-03-25 214 f4bde41dd19db5 Paul E. McKenney 2025-01-08 215 // Converts a per-CPU pointer to an ->srcu_ctrs[] array element to that f4bde41dd19db5 Paul E. McKenney 2025-01-08 216 // element's index. f4bde41dd19db5 Paul E. McKenney 2025-01-08 @217 static inline bool __srcu_ptr_to_ctr(struct srcu_struct *ssp, struct srcu_ctr __percpu *scpp) f4bde41dd19db5 Paul E. McKenney 2025-01-08 218 { f4bde41dd19db5 Paul E. McKenney 2025-01-08 219 return scpp - &ssp->sda->srcu_ctrs[0]; f4bde41dd19db5 Paul E. McKenney 2025-01-08 220 } f4bde41dd19db5 Paul E. McKenney 2025-01-08 221 4937096b579a36 Paul E. McKenney 2025-01-08 222 // Converts an integer to a per-CPU pointer to the corresponding 4937096b579a36 Paul E. McKenney 2025-01-08 223 // ->srcu_ctrs[] array element. 4937096b579a36 Paul E. McKenney 2025-01-08 @224 static inline struct srcu_ctr __percpu *__srcu_ctr_to_ptr(struct srcu_struct *ssp, int idx) 4937096b579a36 Paul E. McKenney 2025-01-08 225 { 4937096b579a36 Paul E. McKenney 2025-01-08 226 return &ssp->sda->srcu_ctrs[idx]; 4937096b579a36 Paul E. McKenney 2025-01-08 227 } 4937096b579a36 Paul E. McKenney 2025-01-08 228 c4020620528e4e Paul E. McKenney 2025-01-09 229 /* c4020620528e4e Paul E. McKenney 2025-01-09 230 * Counts the new reader in the appropriate per-CPU element of the c4020620528e4e Paul E. McKenney 2025-01-09 231 * srcu_struct. Returns a pointer that must be passed to the matching c4020620528e4e Paul E. McKenney 2025-01-09 232 * srcu_read_unlock_fast(). c4020620528e4e Paul E. McKenney 2025-01-09 233 * 3cec27453db49a Paul E. McKenney 2025-01-28 234 * Note that both this_cpu_inc() and atomic_long_inc() are RCU read-side 3cec27453db49a Paul E. McKenney 2025-01-28 235 * critical sections either because they disables interrupts, because they 3cec27453db49a Paul E. McKenney 2025-01-28 236 * are a single instruction, or because they are a read-modify-write atomic 3cec27453db49a Paul E. McKenney 2025-01-28 237 * operation, depending on the whims of the architecture. 3cec27453db49a Paul E. McKenney 2025-01-28 238 * 3cec27453db49a Paul E. McKenney 2025-01-28 239 * This means that __srcu_read_lock_fast() is not all that fast 3cec27453db49a Paul E. McKenney 2025-01-28 240 * on architectures that support NMIs but do not supply NMI-safe 3cec27453db49a Paul E. McKenney 2025-01-28 241 * implementations of this_cpu_inc(). c4020620528e4e Paul E. McKenney 2025-01-09 242 */ c4020620528e4e Paul E. McKenney 2025-01-09 @243 static inline struct srcu_ctr __percpu *__srcu_read_lock_fast(struct srcu_struct *ssp) c4020620528e4e Paul E. McKenney 2025-01-09 244 { c4020620528e4e Paul E. McKenney 2025-01-09 245 struct srcu_ctr __percpu *scp = READ_ONCE(ssp->srcu_ctrp); c4020620528e4e Paul E. McKenney 2025-01-09 246 c4020620528e4e Paul E. McKenney 2025-01-09 247 RCU_LOCKDEP_WARN(!rcu_is_watching(), "RCU must be watching srcu_read_lock_fast()."); 3cec27453db49a Paul E. McKenney 2025-01-28 248 if (!IS_ENABLED(CONFIG_NEED_SRCU_NMI_SAFE)) c4020620528e4e Paul E. McKenney 2025-01-09 249 this_cpu_inc(scp->srcu_locks.counter); /* Y */ 3cec27453db49a Paul E. McKenney 2025-01-28 250 else 3cec27453db49a Paul E. McKenney 2025-01-28 251 atomic_long_inc(raw_cpu_ptr(&scp->srcu_locks)); /* Z */ c4020620528e4e Paul E. McKenney 2025-01-09 252 barrier(); /* Avoid leaking the critical section. */ c4020620528e4e Paul E. McKenney 2025-01-09 253 return scp; c4020620528e4e Paul E. McKenney 2025-01-09 254 } c4020620528e4e Paul E. McKenney 2025-01-09 255 -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki ^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup 2025-04-18 18:12 ` [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup kernel test robot @ 2025-04-22 12:59 ` Uladzislau Rezki 0 siblings, 0 replies; 18+ messages in thread From: Uladzislau Rezki @ 2025-04-22 12:59 UTC (permalink / raw) To: kernel test robot Cc: Andrew Morton, oe-kbuild-all, Linux Memory Management List, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko Hello! Fixed by the test_vmalloc.c: Fix compile error with CONFIG_TINY_RCU -- Vlad Rezki On Fri, Apr 18, 2025 at 8:13 PM kernel test robot <lkp@intel.com> wrote: > > Hi Uladzislau, > > kernel test robot noticed the following build errors: > > [auto build test ERROR on akpm-mm/mm-nonmm-unstable] > [also build test ERROR on linus/master hch-configfs/for-next v6.15-rc2 next-20250417] > [cannot apply to akpm-mm/mm-everything] > [If your patch is applied to the wrong git tree, kindly drop us a note. > And when submitting patch, we suggest to use '--base' as documented in > https://git-scm.com/docs/git-format-patch#_base_tree_information] > > url: https://github.com/intel-lab-lkp/linux/commits/Uladzislau-Rezki-Sony/lib-test_vmalloc-c-Allow-built-in-execution/20250418-001446 > base: https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-nonmm-unstable > patch link: https://lore.kernel.org/r/20250417161216.88318-1-urezki%40gmail.com > patch subject: [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup > config: sparc64-randconfig-001-20250419 (https://download.01.org/0day-ci/archive/20250419/202504190146.gTBIrktk-lkp@intel.com/config) > compiler: sparc64-linux-gcc (GCC) 8.5.0 > reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250419/202504190146.gTBIrktk-lkp@intel.com/reproduce) > > If you fix the issue in a separate patch/commit (i.e. not just a new version of > the same patch/commit), kindly add following tags > | Reported-by: kernel test robot <lkp@intel.com> > | Closes: https://lore.kernel.org/oe-kbuild-all/202504190146.gTBIrktk-lkp@intel.com/ > > All error/warnings (new ones prefixed by >>): > > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:69:8: error: redefinition of 'struct srcu_usage' > struct srcu_usage { > ^~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:52:8: note: originally defined here > struct srcu_usage { }; > ^~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:102:8: error: redefinition of 'struct srcu_struct' > struct srcu_struct { > ^~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:16:8: note: originally defined here > struct srcu_struct { > ^~~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:149: warning: "__SRCU_USAGE_INIT" redefined > #define __SRCU_USAGE_INIT(name) \ > > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:53: note: this is the location of the previous definition > #define __SRCU_USAGE_INIT(name) { } > > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:167: warning: "__SRCU_STRUCT_INIT" redefined > #define __SRCU_STRUCT_INIT(name, usage_name, pcpu_name) \ > > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:34: note: this is the location of the previous definition > #define __SRCU_STRUCT_INIT(name, __ignored, ___ignored) \ > > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:207: warning: "DEFINE_SRCU" redefined > #define DEFINE_SRCU(name) __DEFINE_SRCU(name, /* not static */) > > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:46: note: this is the location of the previous definition > #define DEFINE_SRCU(name) \ > > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:208: warning: "DEFINE_STATIC_SRCU" redefined > #define DEFINE_STATIC_SRCU(name) __DEFINE_SRCU(name, static) > > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:48: note: this is the location of the previous definition > #define DEFINE_STATIC_SRCU(name) \ > > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:210:5: error: conflicting types for '__srcu_read_lock' > int __srcu_read_lock(struct srcu_struct *ssp) __acquires(ssp); > ^~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:63:19: note: previous definition of '__srcu_read_lock' was here > static inline int __srcu_read_lock(struct srcu_struct *ssp) > ^~~~~~~~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:211:6: error: conflicting types for 'synchronize_srcu_expedited' > void synchronize_srcu_expedited(struct srcu_struct *ssp); > ^~~~~~~~~~~~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:99:20: note: previous definition of 'synchronize_srcu_expedited' was here > static inline void synchronize_srcu_expedited(struct srcu_struct *ssp) > ^~~~~~~~~~~~~~~~~~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:212:6: error: conflicting types for 'srcu_barrier' > void srcu_barrier(struct srcu_struct *ssp); > ^~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:104:20: note: previous definition of 'srcu_barrier' was here > static inline void srcu_barrier(struct srcu_struct *ssp) > ^~~~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:213:6: error: conflicting types for 'srcu_torture_stats_print' > void srcu_torture_stats_print(struct srcu_struct *ssp, char *tt, char *tf); > ^~~~~~~~~~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:113:20: note: previous definition of 'srcu_torture_stats_print' was here > static inline void srcu_torture_stats_print(struct srcu_struct *ssp, > ^~~~~~~~~~~~~~~~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:217:20: error: conflicting types for '__srcu_ptr_to_ctr' > static inline bool __srcu_ptr_to_ctr(struct srcu_struct *ssp, struct srcu_ctr __percpu *scpp) > ^~~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:76:20: note: previous definition of '__srcu_ptr_to_ctr' was here > static inline bool __srcu_ptr_to_ctr(struct srcu_struct *ssp, struct srcu_ctr __percpu *scpp) > ^~~~~~~~~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:224:41: error: conflicting types for '__srcu_ctr_to_ptr' > static inline struct srcu_ctr __percpu *__srcu_ctr_to_ptr(struct srcu_struct *ssp, int idx) > ^~~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:81:41: note: previous definition of '__srcu_ctr_to_ptr' was here > static inline struct srcu_ctr __percpu *__srcu_ctr_to_ptr(struct srcu_struct *ssp, int idx) > ^~~~~~~~~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:243:41: error: conflicting types for '__srcu_read_lock_fast' > static inline struct srcu_ctr __percpu *__srcu_read_lock_fast(struct srcu_struct *ssp) > ^~~~~~~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:86:41: note: previous definition of '__srcu_read_lock_fast' was here > static inline struct srcu_ctr __percpu *__srcu_read_lock_fast(struct srcu_struct *ssp) > ^~~~~~~~~~~~~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > >> include/linux/srcutree.h:271:20: error: conflicting types for '__srcu_read_unlock_fast' > static inline void __srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_ctr __percpu *scp) > ^~~~~~~~~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:91:20: note: previous definition of '__srcu_read_unlock_fast' was here > static inline void __srcu_read_unlock_fast(struct srcu_struct *ssp, struct srcu_ctr __percpu *scp) > ^~~~~~~~~~~~~~~~~~~~~~~ > >> include/linux/srcutiny.h:96:31: error: static declaration of '__srcu_read_lock' follows non-static declaration > #define __srcu_read_lock_lite __srcu_read_lock > ^~~~~~~~~~~~~~~~ > include/linux/srcutree.h:291:19: note: in expansion of macro '__srcu_read_lock_lite' > static inline int __srcu_read_lock_lite(struct srcu_struct *ssp) > ^~~~~~~~~~~~~~~~~~~~~ > In file included from lib/test_vmalloc.c:18: > include/linux/srcutree.h:210:5: note: previous declaration of '__srcu_read_lock' was here > int __srcu_read_lock(struct srcu_struct *ssp) __acquires(ssp); > ^~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > >> include/linux/srcutiny.h:97:33: error: conflicting types for '__srcu_read_unlock' > #define __srcu_read_unlock_lite __srcu_read_unlock > ^~~~~~~~~~~~~~~~~~ > include/linux/srcutree.h:312:20: note: in expansion of macro '__srcu_read_unlock_lite' > static inline void __srcu_read_unlock_lite(struct srcu_struct *ssp, int idx) > ^~~~~~~~~~~~~~~~~~~~~~~ > In file included from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcu.h:56:6: note: previous declaration of '__srcu_read_unlock' was here > void __srcu_read_unlock(struct srcu_struct *ssp, int idx) __releases(ssp); > ^~~~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > >> include/linux/srcutiny.h:110:56: error: expected identifier or '(' before 'do' > #define srcu_check_read_flavor_force(ssp, read_flavor) do { } while (0) > ^~ > include/linux/srcutree.h:324:20: note: in expansion of macro 'srcu_check_read_flavor_force' > static inline void srcu_check_read_flavor_force(struct srcu_struct *ssp, int read_flavor) > ^~~~~~~~~~~~~~~~~~~~~~~~~~~~ > >> include/linux/srcutiny.h:110:63: error: expected identifier or '(' before 'while' > #define srcu_check_read_flavor_force(ssp, read_flavor) do { } while (0) > ^~~~~ > include/linux/srcutree.h:324:20: note: in expansion of macro 'srcu_check_read_flavor_force' > static inline void srcu_check_read_flavor_force(struct srcu_struct *ssp, int read_flavor) > ^~~~~~~~~~~~~~~~~~~~~~~~~~~~ > include/linux/srcutiny.h:109:50: error: expected identifier or '(' before 'do' > #define srcu_check_read_flavor(ssp, read_flavor) do { } while (0) > ^~ > include/linux/srcutree.h:336:20: note: in expansion of macro 'srcu_check_read_flavor' > static inline void srcu_check_read_flavor(struct srcu_struct *ssp, int read_flavor) > ^~~~~~~~~~~~~~~~~~~~~~ > include/linux/srcutiny.h:109:57: error: expected identifier or '(' before 'while' > #define srcu_check_read_flavor(ssp, read_flavor) do { } while (0) > ^~~~~ > include/linux/srcutree.h:336:20: note: in expansion of macro 'srcu_check_read_flavor' > static inline void srcu_check_read_flavor(struct srcu_struct *ssp, int read_flavor) > ^~~~~~~~~~~~~~~~~~~~~~ > lib/test_vmalloc.c: In function 'test_func': > >> lib/test_vmalloc.c:460:19: error: passing argument 1 of 'synchronize_srcu' from incompatible pointer type [-Werror=incompatible-pointer-types] > synchronize_srcu(&prepare_for_test_srcu); > ^~~~~~~~~~~~~~~~~~~~~~ > In file included from include/linux/srcu.h:59, > from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcutiny.h:55:43: note: expected 'struct srcu_struct *' but argument is of type 'struct srcu_struct *' > void synchronize_srcu(struct srcu_struct *ssp); > ~~~~~~~~~~~~~~~~~~~~^~~ > lib/test_vmalloc.c: In function 'do_concurrent_test': > >> lib/test_vmalloc.c:538:23: error: passing argument 1 of 'srcu_read_lock' from incompatible pointer type [-Werror=incompatible-pointer-types] > idx = srcu_read_lock(&prepare_for_test_srcu); > ^~~~~~~~~~~~~~~~~~~~~~ > In file included from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcu.h:247:54: note: expected 'struct srcu_struct *' but argument is of type 'struct srcu_struct *' > static inline int srcu_read_lock(struct srcu_struct *ssp) __acquires(ssp) > ~~~~~~~~~~~~~~~~~~~~^~~ > lib/test_vmalloc.c:555:19: error: passing argument 1 of 'srcu_read_unlock' from incompatible pointer type [-Werror=incompatible-pointer-types] > srcu_read_unlock(&prepare_for_test_srcu, idx); > ^~~~~~~~~~~~~~~~~~~~~~ > In file included from include/linux/notifier.h:16, > from include/linux/memory_hotplug.h:7, > from include/linux/mmzone.h:1494, > from include/linux/gfp.h:7, > from include/linux/umh.h:4, > from include/linux/kmod.h:9, > from include/linux/module.h:17, > from lib/test_vmalloc.c:9: > include/linux/srcu.h:397:57: note: expected 'struct srcu_struct *' but argument is of type 'struct srcu_struct *' > static inline void srcu_read_unlock(struct srcu_struct *ssp, int idx) > ~~~~~~~~~~~~~~~~~~~~^~~ > cc1: some warnings being treated as errors > .. > > > vim +69 include/linux/srcutree.h > > da915ad5cf25b5 Paul E. McKenney 2017-04-05 65 > da915ad5cf25b5 Paul E. McKenney 2017-04-05 66 /* > 95433f7263011e Paul E. McKenney 2023-03-16 67 * Per-SRCU-domain structure, update-side data linked from srcu_struct. > da915ad5cf25b5 Paul E. McKenney 2017-04-05 68 */ > 95433f7263011e Paul E. McKenney 2023-03-16 @69 struct srcu_usage { > 2ec303113d9789 Paul E. McKenney 2022-01-21 70 struct srcu_node *node; /* Combining tree. */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 71 struct srcu_node *level[RCU_NUM_LVLS + 1]; > da915ad5cf25b5 Paul E. McKenney 2017-04-05 72 /* First node at each level. */ > 994f706872e6ce Paul E. McKenney 2022-01-24 73 int srcu_size_state; /* Small-to-big transition state. */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 74 struct mutex srcu_cb_mutex; /* Serialize CB preparation. */ > 994f706872e6ce Paul E. McKenney 2022-01-24 75 spinlock_t __private lock; /* Protect counters and size state. */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 76 struct mutex srcu_gp_mutex; /* Serialize GP work. */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 77 unsigned long srcu_gp_seq; /* Grace-period seq #. */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 78 unsigned long srcu_gp_seq_needed; /* Latest gp_seq needed. */ > 1e9a038b7fe9a8 Paul E. McKenney 2017-04-24 79 unsigned long srcu_gp_seq_needed_exp; /* Furthest future exp GP. */ > 282d8998e9979c Paul E. McKenney 2022-03-08 80 unsigned long srcu_gp_start; /* Last GP start timestamp (jiffies) */ > 22607d66bbc3e8 Paul E. McKenney 2017-04-25 81 unsigned long srcu_last_gp_end; /* Last GP end timestamp (ns) */ > 9f2e91d94c9155 Paul E. McKenney 2022-01-27 82 unsigned long srcu_size_jiffies; /* Current contention-measurement interval. */ > 9f2e91d94c9155 Paul E. McKenney 2022-01-27 83 unsigned long srcu_n_lock_retries; /* Contention events in current interval. */ > 282d8998e9979c Paul E. McKenney 2022-03-08 84 unsigned long srcu_n_exp_nodelay; /* # expedited no-delays in current GP phase. */ > 46470cf85d2b61 Paul E. McKenney 2022-01-27 85 bool sda_is_static; /* May ->sda be passed to free_percpu()? */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 86 unsigned long srcu_barrier_seq; /* srcu_barrier seq #. */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 87 struct mutex srcu_barrier_mutex; /* Serialize barrier ops. */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 88 struct completion srcu_barrier_completion; > da915ad5cf25b5 Paul E. McKenney 2017-04-05 89 /* Awaken barrier rq at end. */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 90 atomic_t srcu_barrier_cpu_cnt; /* # CPUs not yet posting a */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 91 /* callback for the barrier */ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 92 /* operation. */ > 282d8998e9979c Paul E. McKenney 2022-03-08 93 unsigned long reschedule_jiffies; > 282d8998e9979c Paul E. McKenney 2022-03-08 94 unsigned long reschedule_count; > d8be81735aa894 Paul E. McKenney 2017-03-25 95 struct delayed_work work; > fd1b3f8e097b7f Paul E. McKenney 2023-03-17 96 struct srcu_struct *srcu_ssp; > 95433f7263011e Paul E. McKenney 2023-03-16 97 }; > 95433f7263011e Paul E. McKenney 2023-03-16 98 > da915ad5cf25b5 Paul E. McKenney 2017-04-05 99 /* > da915ad5cf25b5 Paul E. McKenney 2017-04-05 100 * Per-SRCU-domain structure, similar in function to rcu_state. > da915ad5cf25b5 Paul E. McKenney 2017-04-05 101 */ > d8be81735aa894 Paul E. McKenney 2017-03-25 @102 struct srcu_struct { > 795e7efec6ea7e Paul E. McKenney 2024-12-19 103 struct srcu_ctr __percpu *srcu_ctrp; > da915ad5cf25b5 Paul E. McKenney 2017-04-05 104 struct srcu_data __percpu *sda; /* Per-CPU srcu_data array. */ > d8be81735aa894 Paul E. McKenney 2017-03-25 105 struct lockdep_map dep_map; > 95433f7263011e Paul E. McKenney 2023-03-16 106 struct srcu_usage *srcu_sup; /* Update-side data. */ > d8be81735aa894 Paul E. McKenney 2017-03-25 107 }; > d8be81735aa894 Paul E. McKenney 2017-03-25 108 > e15a19306004b3 Pingfan Liu 2023-01-04 109 // Values for size state variable (->srcu_size_state). Once the state > e15a19306004b3 Pingfan Liu 2023-01-04 110 // has been set to SRCU_SIZE_ALLOC, the grace-period code advances through > e15a19306004b3 Pingfan Liu 2023-01-04 111 // this state machine one step per grace period until the SRCU_SIZE_BIG state > e15a19306004b3 Pingfan Liu 2023-01-04 112 // is reached. Otherwise, the state machine remains in the SRCU_SIZE_SMALL > e15a19306004b3 Pingfan Liu 2023-01-04 113 // state indefinitely. > e15a19306004b3 Pingfan Liu 2023-01-04 114 #define SRCU_SIZE_SMALL 0 // No srcu_node combining tree, ->node == NULL > e15a19306004b3 Pingfan Liu 2023-01-04 115 #define SRCU_SIZE_ALLOC 1 // An srcu_node tree is being allocated, initialized, > e15a19306004b3 Pingfan Liu 2023-01-04 116 // and then referenced by ->node. It will not be used. > e15a19306004b3 Pingfan Liu 2023-01-04 117 #define SRCU_SIZE_WAIT_BARRIER 2 // The srcu_node tree starts being used by everything > e15a19306004b3 Pingfan Liu 2023-01-04 118 // except call_srcu(), especially by srcu_barrier(). > e15a19306004b3 Pingfan Liu 2023-01-04 119 // By the end of this state, all CPUs and threads > e15a19306004b3 Pingfan Liu 2023-01-04 120 // are aware of this tree's existence. > e15a19306004b3 Pingfan Liu 2023-01-04 121 #define SRCU_SIZE_WAIT_CALL 3 // The srcu_node tree starts being used by call_srcu(). > e15a19306004b3 Pingfan Liu 2023-01-04 122 // By the end of this state, all of the call_srcu() > e15a19306004b3 Pingfan Liu 2023-01-04 123 // invocations that were running on a non-boot CPU > e15a19306004b3 Pingfan Liu 2023-01-04 124 // and using the boot CPU's callback queue will have > e15a19306004b3 Pingfan Liu 2023-01-04 125 // completed. > e15a19306004b3 Pingfan Liu 2023-01-04 126 #define SRCU_SIZE_WAIT_CBS1 4 // Don't trust the ->srcu_have_cbs[] grace-period > e15a19306004b3 Pingfan Liu 2023-01-04 127 #define SRCU_SIZE_WAIT_CBS2 5 // sequence elements or the ->srcu_data_have_cbs[] > e15a19306004b3 Pingfan Liu 2023-01-04 128 #define SRCU_SIZE_WAIT_CBS3 6 // CPU-bitmask elements until all four elements of > e15a19306004b3 Pingfan Liu 2023-01-04 129 #define SRCU_SIZE_WAIT_CBS4 7 // each array have been initialized. > e15a19306004b3 Pingfan Liu 2023-01-04 130 #define SRCU_SIZE_BIG 8 // The srcu_node combining tree is fully initialized > e15a19306004b3 Pingfan Liu 2023-01-04 131 // and all aspects of it are being put to use. > 994f706872e6ce Paul E. McKenney 2022-01-24 132 > da915ad5cf25b5 Paul E. McKenney 2017-04-05 133 /* Values for state variable (bottom bits of ->srcu_gp_seq). */ > d8be81735aa894 Paul E. McKenney 2017-03-25 134 #define SRCU_STATE_IDLE 0 > d8be81735aa894 Paul E. McKenney 2017-03-25 135 #define SRCU_STATE_SCAN1 1 > d8be81735aa894 Paul E. McKenney 2017-03-25 136 #define SRCU_STATE_SCAN2 2 > d8be81735aa894 Paul E. McKenney 2017-03-25 137 > 29bc83e4d90546 JP Kobryn 2024-07-15 138 /* > 29bc83e4d90546 JP Kobryn 2024-07-15 139 * Values for initializing gp sequence fields. Higher values allow wrap arounds to > 29bc83e4d90546 JP Kobryn 2024-07-15 140 * occur earlier. > 29bc83e4d90546 JP Kobryn 2024-07-15 141 * The second value with state is useful in the case of static initialization of > 29bc83e4d90546 JP Kobryn 2024-07-15 142 * srcu_usage where srcu_gp_seq_needed is expected to have some state value in its > 29bc83e4d90546 JP Kobryn 2024-07-15 143 * lower bits (or else it will appear to be already initialized within > 29bc83e4d90546 JP Kobryn 2024-07-15 144 * the call check_init_srcu_struct()). > 29bc83e4d90546 JP Kobryn 2024-07-15 145 */ > 29bc83e4d90546 JP Kobryn 2024-07-15 146 #define SRCU_GP_SEQ_INITIAL_VAL ((0UL - 100UL) << RCU_SEQ_CTR_SHIFT) > 29bc83e4d90546 JP Kobryn 2024-07-15 147 #define SRCU_GP_SEQ_INITIAL_VAL_WITH_STATE (SRCU_GP_SEQ_INITIAL_VAL - 1) > 29bc83e4d90546 JP Kobryn 2024-07-15 148 > 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 @149 #define __SRCU_USAGE_INIT(name) \ > d8be81735aa894 Paul E. McKenney 2017-03-25 150 { \ > d633198088bd9e Paul E. McKenney 2017-10-10 151 .lock = __SPIN_LOCK_UNLOCKED(name.lock), \ > 29bc83e4d90546 JP Kobryn 2024-07-15 152 .srcu_gp_seq = SRCU_GP_SEQ_INITIAL_VAL, \ > 29bc83e4d90546 JP Kobryn 2024-07-15 153 .srcu_gp_seq_needed = SRCU_GP_SEQ_INITIAL_VAL_WITH_STATE, \ > 29bc83e4d90546 JP Kobryn 2024-07-15 154 .srcu_gp_seq_needed_exp = SRCU_GP_SEQ_INITIAL_VAL, \ > 4e6ea4ef56f942 Paul E. McKenney 2018-08-14 155 .work = __DELAYED_WORK_INITIALIZER(name.work, NULL, 0), \ > 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 156 } > 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 157 > 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 158 #define __SRCU_STRUCT_INIT_COMMON(name, usage_name) \ > 95433f7263011e Paul E. McKenney 2023-03-16 159 .srcu_sup = &usage_name, \ > f4d01a259374ef Paul E. McKenney 2023-03-17 160 __SRCU_DEP_MAP_INIT(name) > f4d01a259374ef Paul E. McKenney 2023-03-17 161 > 95433f7263011e Paul E. McKenney 2023-03-16 162 #define __SRCU_STRUCT_INIT_MODULE(name, usage_name) \ > f4d01a259374ef Paul E. McKenney 2023-03-17 163 { \ > 95433f7263011e Paul E. McKenney 2023-03-16 164 __SRCU_STRUCT_INIT_COMMON(name, usage_name) \ > d8be81735aa894 Paul E. McKenney 2017-03-25 165 } > d8be81735aa894 Paul E. McKenney 2017-03-25 166 > 95433f7263011e Paul E. McKenney 2023-03-16 @167 #define __SRCU_STRUCT_INIT(name, usage_name, pcpu_name) \ > f4d01a259374ef Paul E. McKenney 2023-03-17 168 { \ > f4d01a259374ef Paul E. McKenney 2023-03-17 169 .sda = &pcpu_name, \ > 795e7efec6ea7e Paul E. McKenney 2024-12-19 170 .srcu_ctrp = &pcpu_name.srcu_ctrs[0], \ > 95433f7263011e Paul E. McKenney 2023-03-16 171 __SRCU_STRUCT_INIT_COMMON(name, usage_name) \ > d8be81735aa894 Paul E. McKenney 2017-03-25 172 } > d8be81735aa894 Paul E. McKenney 2017-03-25 173 > d8be81735aa894 Paul E. McKenney 2017-03-25 174 /* > d8be81735aa894 Paul E. McKenney 2017-03-25 175 * Define and initialize a srcu struct at build time. > d8be81735aa894 Paul E. McKenney 2017-03-25 176 * Do -not- call init_srcu_struct() nor cleanup_srcu_struct() on it. > d8be81735aa894 Paul E. McKenney 2017-03-25 177 * > d8be81735aa894 Paul E. McKenney 2017-03-25 178 * Note that although DEFINE_STATIC_SRCU() hides the name from other > d8be81735aa894 Paul E. McKenney 2017-03-25 179 * files, the per-CPU variable rules nevertheless require that the > d8be81735aa894 Paul E. McKenney 2017-03-25 180 * chosen name be globally unique. These rules also prohibit use of > d8be81735aa894 Paul E. McKenney 2017-03-25 181 * DEFINE_STATIC_SRCU() within a function. If these rules are too > d8be81735aa894 Paul E. McKenney 2017-03-25 182 * restrictive, declare the srcu_struct manually. For example, in > d8be81735aa894 Paul E. McKenney 2017-03-25 183 * each file: > d8be81735aa894 Paul E. McKenney 2017-03-25 184 * > d8be81735aa894 Paul E. McKenney 2017-03-25 185 * static struct srcu_struct my_srcu; > d8be81735aa894 Paul E. McKenney 2017-03-25 186 * > d8be81735aa894 Paul E. McKenney 2017-03-25 187 * Then, before the first use of each my_srcu, manually initialize it: > d8be81735aa894 Paul E. McKenney 2017-03-25 188 * > d8be81735aa894 Paul E. McKenney 2017-03-25 189 * init_srcu_struct(&my_srcu); > d8be81735aa894 Paul E. McKenney 2017-03-25 190 * > d8be81735aa894 Paul E. McKenney 2017-03-25 191 * See include/linux/percpu-defs.h for the rules on per-CPU variables. > d8be81735aa894 Paul E. McKenney 2017-03-25 192 */ > fe15b50cdeeebd Paul E. McKenney 2019-04-05 193 #ifdef MODULE > fe15b50cdeeebd Paul E. McKenney 2019-04-05 194 # define __DEFINE_SRCU(name, is_static) \ > 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 195 static struct srcu_usage name##_srcu_usage = __SRCU_USAGE_INIT(name##_srcu_usage); \ > 95433f7263011e Paul E. McKenney 2023-03-16 196 is_static struct srcu_struct name = __SRCU_STRUCT_INIT_MODULE(name, name##_srcu_usage); \ > db8f1471c61336 Alexander Aring 2022-01-26 197 extern struct srcu_struct * const __srcu_struct_##name; \ > 056b89e7e69974 Joel Fernandes (Google 2019-04-11 198) struct srcu_struct * const __srcu_struct_##name \ > fe15b50cdeeebd Paul E. McKenney 2019-04-05 199 __section("___srcu_struct_ptrs") = &name > fe15b50cdeeebd Paul E. McKenney 2019-04-05 200 #else > d8be81735aa894 Paul E. McKenney 2017-03-25 201 # define __DEFINE_SRCU(name, is_static) \ > da915ad5cf25b5 Paul E. McKenney 2017-04-05 202 static DEFINE_PER_CPU(struct srcu_data, name##_srcu_data); \ > 03200b5ca3b4d4 Paul E. McKenney 2023-03-17 203 static struct srcu_usage name##_srcu_usage = __SRCU_USAGE_INIT(name##_srcu_usage); \ > fe15b50cdeeebd Paul E. McKenney 2019-04-05 204 is_static struct srcu_struct name = \ > 95433f7263011e Paul E. McKenney 2023-03-16 205 __SRCU_STRUCT_INIT(name, name##_srcu_usage, name##_srcu_data) > fe15b50cdeeebd Paul E. McKenney 2019-04-05 206 #endif > d8be81735aa894 Paul E. McKenney 2017-03-25 @207 #define DEFINE_SRCU(name) __DEFINE_SRCU(name, /* not static */) > d8be81735aa894 Paul E. McKenney 2017-03-25 @208 #define DEFINE_STATIC_SRCU(name) __DEFINE_SRCU(name, static) > d8be81735aa894 Paul E. McKenney 2017-03-25 209 > 443971156cebfc Paul E. McKenney 2025-01-13 @210 int __srcu_read_lock(struct srcu_struct *ssp) __acquires(ssp); > aacb5d91ab1bfb Paul E. McKenney 2018-10-28 @211 void synchronize_srcu_expedited(struct srcu_struct *ssp); > aacb5d91ab1bfb Paul E. McKenney 2018-10-28 @212 void srcu_barrier(struct srcu_struct *ssp); > aacb5d91ab1bfb Paul E. McKenney 2018-10-28 @213 void srcu_torture_stats_print(struct srcu_struct *ssp, char *tt, char *tf); > d8be81735aa894 Paul E. McKenney 2017-03-25 214 > f4bde41dd19db5 Paul E. McKenney 2025-01-08 215 // Converts a per-CPU pointer to an ->srcu_ctrs[] array element to that > f4bde41dd19db5 Paul E. McKenney 2025-01-08 216 // element's index. > f4bde41dd19db5 Paul E. McKenney 2025-01-08 @217 static inline bool __srcu_ptr_to_ctr(struct srcu_struct *ssp, struct srcu_ctr __percpu *scpp) > f4bde41dd19db5 Paul E. McKenney 2025-01-08 218 { > f4bde41dd19db5 Paul E. McKenney 2025-01-08 219 return scpp - &ssp->sda->srcu_ctrs[0]; > f4bde41dd19db5 Paul E. McKenney 2025-01-08 220 } > f4bde41dd19db5 Paul E. McKenney 2025-01-08 221 > 4937096b579a36 Paul E. McKenney 2025-01-08 222 // Converts an integer to a per-CPU pointer to the corresponding > 4937096b579a36 Paul E. McKenney 2025-01-08 223 // ->srcu_ctrs[] array element. > 4937096b579a36 Paul E. McKenney 2025-01-08 @224 static inline struct srcu_ctr __percpu *__srcu_ctr_to_ptr(struct srcu_struct *ssp, int idx) > 4937096b579a36 Paul E. McKenney 2025-01-08 225 { > 4937096b579a36 Paul E. McKenney 2025-01-08 226 return &ssp->sda->srcu_ctrs[idx]; > 4937096b579a36 Paul E. McKenney 2025-01-08 227 } > 4937096b579a36 Paul E. McKenney 2025-01-08 228 > c4020620528e4e Paul E. McKenney 2025-01-09 229 /* > c4020620528e4e Paul E. McKenney 2025-01-09 230 * Counts the new reader in the appropriate per-CPU element of the > c4020620528e4e Paul E. McKenney 2025-01-09 231 * srcu_struct. Returns a pointer that must be passed to the matching > c4020620528e4e Paul E. McKenney 2025-01-09 232 * srcu_read_unlock_fast(). > c4020620528e4e Paul E. McKenney 2025-01-09 233 * > 3cec27453db49a Paul E. McKenney 2025-01-28 234 * Note that both this_cpu_inc() and atomic_long_inc() are RCU read-side > 3cec27453db49a Paul E. McKenney 2025-01-28 235 * critical sections either because they disables interrupts, because they > 3cec27453db49a Paul E. McKenney 2025-01-28 236 * are a single instruction, or because they are a read-modify-write atomic > 3cec27453db49a Paul E. McKenney 2025-01-28 237 * operation, depending on the whims of the architecture. > 3cec27453db49a Paul E. McKenney 2025-01-28 238 * > 3cec27453db49a Paul E. McKenney 2025-01-28 239 * This means that __srcu_read_lock_fast() is not all that fast > 3cec27453db49a Paul E. McKenney 2025-01-28 240 * on architectures that support NMIs but do not supply NMI-safe > 3cec27453db49a Paul E. McKenney 2025-01-28 241 * implementations of this_cpu_inc(). > c4020620528e4e Paul E. McKenney 2025-01-09 242 */ > c4020620528e4e Paul E. McKenney 2025-01-09 @243 static inline struct srcu_ctr __percpu *__srcu_read_lock_fast(struct srcu_struct *ssp) > c4020620528e4e Paul E. McKenney 2025-01-09 244 { > c4020620528e4e Paul E. McKenney 2025-01-09 245 struct srcu_ctr __percpu *scp = READ_ONCE(ssp->srcu_ctrp); > c4020620528e4e Paul E. McKenney 2025-01-09 246 > c4020620528e4e Paul E. McKenney 2025-01-09 247 RCU_LOCKDEP_WARN(!rcu_is_watching(), "RCU must be watching srcu_read_lock_fast()."); > 3cec27453db49a Paul E. McKenney 2025-01-28 248 if (!IS_ENABLED(CONFIG_NEED_SRCU_NMI_SAFE)) > c4020620528e4e Paul E. McKenney 2025-01-09 249 this_cpu_inc(scp->srcu_locks.counter); /* Y */ > 3cec27453db49a Paul E. McKenney 2025-01-28 250 else > 3cec27453db49a Paul E. McKenney 2025-01-28 251 atomic_long_inc(raw_cpu_ptr(&scp->srcu_locks)); /* Z */ > c4020620528e4e Paul E. McKenney 2025-01-09 252 barrier(); /* Avoid leaking the critical section. */ > c4020620528e4e Paul E. McKenney 2025-01-09 253 return scp; > c4020620528e4e Paul E. McKenney 2025-01-09 254 } > c4020620528e4e Paul E. McKenney 2025-01-09 255 > > -- > 0-DAY CI Kernel Test Service > https://github.com/intel/lkp-tests/wiki -- Uladzislau Rezki ^ permalink raw reply [flat|nested] 18+ messages in thread
* RE: [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup 2025-04-17 16:12 [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup Uladzislau Rezki (Sony) ` (3 preceding siblings ...) 2025-04-18 18:12 ` [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup kernel test robot @ 2025-04-24 13:35 ` Adrian Huang12 4 siblings, 0 replies; 18+ messages in thread From: Adrian Huang12 @ 2025-04-24 13:35 UTC (permalink / raw) To: Uladzislau Rezki (Sony), Andrew Morton Cc: linux-mm@kvack.org, LKML, Baoquan He, Christoph Hellwig, Oleksiy Avramchenko > -----Original Message----- > From: owner-linux-mm@kvack.org <owner-linux-mm@kvack.org> On Behalf > Of Uladzislau Rezki (Sony) > Sent: Friday, April 18, 2025 12:12 AM > To: Andrew Morton <akpm@linux-foundation.org> > Cc: linux-mm@kvack.org; LKML <linux-kernel@vger.kernel.org>; Baoquan He > <bhe@redhat.com>; Christoph Hellwig <hch@infradead.org>; Uladzislau Rezki > <urezki@gmail.com>; Oleksiy Avramchenko <oleksiy.avramchenko@sony.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@gmail.com> > --- > lib/test_vmalloc.c | 17 +++++++---------- > 1 file changed, 7 insertions(+), 10 deletions(-) Reviewed-by: Adrian Huang <ahuang12@lenovo.com> Tested-by: Adrian Huang <ahuang12@lenovo.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 > ^ permalink raw reply [flat|nested] 18+ messages in thread
end of thread, other threads:[~2025-06-14 15:35 UTC | newest] Thread overview: 18+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2025-04-17 16:12 [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup Uladzislau Rezki (Sony) 2025-04-17 16:12 ` [PATCH 2/4] lib/test_vmalloc.c: Allow built-in execution Uladzislau Rezki (Sony) 2025-04-18 15:01 ` Baoquan He 2025-04-24 13:37 ` Adrian Huang12 2025-06-13 13:59 ` Dev Jain 2025-06-13 14:19 ` Uladzislau Rezki 2025-06-13 14:23 ` Dev Jain 2025-06-13 14:37 ` Uladzislau Rezki 2025-06-14 10:27 ` Dev Jain 2025-06-14 15:35 ` Uladzislau Rezki 2025-04-17 16:12 ` [PATCH 3/4] MAINTAINERS: Add test_vmalloc.c to VMALLOC section Uladzislau Rezki (Sony) 2025-04-18 15:01 ` Baoquan He 2025-04-17 16:12 ` [PATCH 4/4] vmalloc: Align nr_vmalloc_pages and vmap_lazy_nr Uladzislau Rezki (Sony) 2025-04-18 13:26 ` Baoquan He 2025-04-24 13:38 ` Adrian Huang12 2025-04-18 18:12 ` [PATCH 1/4] lib/test_vmalloc.c: Replace RWSEM to SRCU for setup kernel test robot 2025-04-22 12:59 ` Uladzislau Rezki 2025-04-24 13:35 ` Adrian Huang12
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).