* Crash when memset of shared mapped memory in ARM
@ 2011-11-16 5:35 naveen yadav
2011-11-16 8:27 ` Russell King - ARM Linux
2011-11-16 8:45 ` Russell King - ARM Linux
0 siblings, 2 replies; 6+ messages in thread
From: naveen yadav @ 2011-11-16 5:35 UTC (permalink / raw)
To: Russell King - ARM Linux, linux-mm, linux-arm-kernel-request,
linux-arm-kernel, linux-arm-request, linux-kernel
Hi All,
I am running below Test program on ARM cortex a9/a8 on kernel version
2.6.35.14 as well as on 3.0.
Please find the test case where:
1. Create shared memory object using shm_open(If we use normal open
then no problem only problem with shm_open)
2. ftruncate to given size
3. memory map the shared object to given memory address ( I haved
tested without MAP_SHARED, MAP_FIXED as well, problem exist)
4. Memset the shared memory (got page fault when accessing the second page)
Observation: Only observed in ARM ( i.e not present in MIPS and X86)
#undef NDEBUG
#define _GNU_SOURCE
#include <unistd.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/ipc.h>
#include <sys/shm.h>
#include <errno.h>
#include <pthread.h>
#include <string.h>
#include <signal.h>
#include <fcntl.h>
#include <sys/types.h>
#include <sys/mman.h>
enum {
SHM_INIT,
SHM_GET
};
enum {
PARENT,
CHILD
};
#define FIXED_MMAP_ADDR 0x20000000
#define MMAP_SIZE 0x10000
static int shmid;
static char shm_name[100];
static int sleep_period = 100000;
void * shmem_init(int flag)
{
int start = FIXED_MMAP_ADDR;
int memory_size = MMAP_SIZE;
int mode = 0666;
void *addr;
int ret;
sprintf(shm_name, "/shmem_1234");
shmid = shm_open (shm_name, O_RDWR | O_EXCL | O_CREAT | O_TRUNC, mode);
if (shmid < 0) {
if (errno == EEXIST) {
printf ("shm_open: %s\n", strerror(errno));
shmid = shm_open (shm_name, O_RDWR, mode);
} else {
printf("failed to shm_open, err=%s\n", strerror(errno));
return NULL;
}
}
ret = fcntl (shmid, F_SETFD, FD_CLOEXEC);
if (ret < 0) {
printf("fcntl: %s\n", strerror(errno));
return NULL;
}
ret = ftruncate (shmid, memory_size);
if (ret < 0) {
printf("ftruncate: %s\n", strerror(errno));
return NULL;
}
addr = mmap ((void *)start, memory_size, PROT_READ | PROT_WRITE,
MAP_SHARED | MAP_FIXED, shmid, 0);
if (addr == MAP_FAILED) {
printf ("mmap: %s\n", strerror(errno));
close (shmid);
shm_unlink (shm_name);
return NULL;
}
if (flag == SHM_INIT){
printf ("mmap: addr %p\n", addr);
/* memset on arm creates a unhandled page fault, works fine on mips */
memset(addr, 0, memory_size);
}
return (void *)addr;
}
pthread_mutex_t * shmem_mutex_init(int flag)
{
pthread_mutex_t * pmutex = (pthread_mutex_t *)shmem_init(flag);
#if 0
pthread_mutexattr_t attr;
if (flag == SHM_INIT) {
pthread_mutexattr_init (&attr);
pthread_mutexattr_setpshared (&attr, PTHREAD_PROCESS_SHARED);
pthread_mutexattr_setprotocol (&attr, PTHREAD_PRIO_INHERIT);
pthread_mutexattr_setrobust_np (&attr,
PTHREAD_MUTEX_STALLED_NP);
pthread_mutexattr_settype (&attr, PTHREAD_MUTEX_ERRORCHECK);
if (pthread_mutex_init (pmutex, &attr) != 0) {
printf("Init mutex failed, err=%s\n", strerror(errno));
pthread_mutexattr_destroy (&attr);
return NULL;
}
}
#endif
return pmutex;
}
void long_running_task(int flag)
{
static int counter = 0;
if (flag == PARENT)
usleep(5*sleep_period);
else
usleep(3*sleep_period);
counter = (counter + 1) % 100;
printf("%d: completed %d computing\n", getpid(), counter);
}
void sig_handler(int signum)
{
close(shmid);
shm_unlink(shm_name);
exit(0);
}
int main(int argc, char *argv[])
{
pthread_mutex_t *mutex_parent, *mutex_child;
// signal(SIGUSR1, sig_handler);
// if (fork()) {
/* parent process */
if ((mutex_parent = shmem_mutex_init(SHM_INIT)) == NULL) {
printf("failed to get the shmem_mutex\n");
exit(-1);
}
#if 0
while (1) {
printf("%d: try to hold the lock\n", getpid());
pthread_mutex_lock(mutex_parent);
printf("%d: got the lock\n", getpid());
long_running_task(PARENT);
pthread_mutex_unlock(mutex_parent);
printf("%d: released the lock\n", getpid());
}
#endif
// } else {
#if 0
/* child process */
usleep(sleep_period);
if ((mutex_child = shmem_mutex_init(SHM_GET)) == NULL) {
printf("failed to get the shmem_mutex\n");
exit(-1);
}
while (1) {
printf("%d: try to hold the lock\n", getpid());
pthread_mutex_lock(mutex_child);
printf("%d: got the lock\n", getpid());
long_running_task(CHILD);
pthread_mutex_unlock(mutex_child);
printf("%d: released the lock\n", getpid());
}
#endif
// }
return 0;
}
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Crash when memset of shared mapped memory in ARM
2011-11-16 5:35 Crash when memset of shared mapped memory in ARM naveen yadav
@ 2011-11-16 8:27 ` Russell King - ARM Linux
2011-11-16 9:07 ` naveen yadav
2011-11-16 8:45 ` Russell King - ARM Linux
1 sibling, 1 reply; 6+ messages in thread
From: Russell King - ARM Linux @ 2011-11-16 8:27 UTC (permalink / raw)
To: naveen yadav
Cc: linux-mm, linux-arm-kernel-request, linux-arm-kernel,
linux-arm-request, linux-kernel
On Wed, Nov 16, 2011 at 11:05:49AM +0530, naveen yadav wrote:
> Hi All,
>
> I am running below Test program on ARM cortex a9/a8 on kernel version
> 2.6.35.14 as well as on 3.0.
>
> Please find the test case where:
>
> 1. Create shared memory object using shm_open(If we use normal open
> then no problem only problem with shm_open)
>
> 2. ftruncate to given size
>
> 3. memory map the shared object to given memory address ( I haved
> tested without MAP_SHARED, MAP_FIXED as well, problem exist)
>
> 4. Memset the shared memory (got page fault when accessing the second page)
>
>
>
>
> Observation: Only observed in ARM ( i.e not present in MIPS and X86)
Yes, so, what happens?
>
>
> #undef NDEBUG
> #define _GNU_SOURCE
> #include <unistd.h>
> #include <stdio.h>
> #include <stdlib.h>
> #include <sys/ipc.h>
> #include <sys/shm.h>
> #include <errno.h>
> #include <pthread.h>
> #include <string.h>
> #include <signal.h>
> #include <fcntl.h>
> #include <sys/types.h>
> #include <sys/mman.h>
>
> enum {
> SHM_INIT,
> SHM_GET
> };
>
> enum {
> PARENT,
> CHILD
> };
>
> #define FIXED_MMAP_ADDR 0x20000000
> #define MMAP_SIZE 0x10000
>
> static int shmid;
> static char shm_name[100];
> static int sleep_period = 100000;
> void * shmem_init(int flag)
> {
> int start = FIXED_MMAP_ADDR;
> int memory_size = MMAP_SIZE;
> int mode = 0666;
> void *addr;
> int ret;
> sprintf(shm_name, "/shmem_1234");
> shmid = shm_open (shm_name, O_RDWR | O_EXCL | O_CREAT | O_TRUNC, mode);
> if (shmid < 0) {
> if (errno == EEXIST) {
> printf ("shm_open: %s\n", strerror(errno));
> shmid = shm_open (shm_name, O_RDWR, mode);
>
> } else {
> printf("failed to shm_open, err=%s\n", strerror(errno));
> return NULL;
> }
> }
> ret = fcntl (shmid, F_SETFD, FD_CLOEXEC);
> if (ret < 0) {
> printf("fcntl: %s\n", strerror(errno));
> return NULL;
> }
> ret = ftruncate (shmid, memory_size);
> if (ret < 0) {
> printf("ftruncate: %s\n", strerror(errno));
> return NULL;
> }
> addr = mmap ((void *)start, memory_size, PROT_READ | PROT_WRITE,
> MAP_SHARED | MAP_FIXED, shmid, 0);
> if (addr == MAP_FAILED) {
> printf ("mmap: %s\n", strerror(errno));
> close (shmid);
> shm_unlink (shm_name);
> return NULL;
> }
>
> if (flag == SHM_INIT){
> printf ("mmap: addr %p\n", addr);
> /* memset on arm creates a unhandled page fault, works fine on mips */
> memset(addr, 0, memory_size);
> }
> return (void *)addr;
> }
>
> pthread_mutex_t * shmem_mutex_init(int flag)
> {
> pthread_mutex_t * pmutex = (pthread_mutex_t *)shmem_init(flag);
> #if 0
> pthread_mutexattr_t attr;
> if (flag == SHM_INIT) {
> pthread_mutexattr_init (&attr);
> pthread_mutexattr_setpshared (&attr, PTHREAD_PROCESS_SHARED);
> pthread_mutexattr_setprotocol (&attr, PTHREAD_PRIO_INHERIT);
> pthread_mutexattr_setrobust_np (&attr,
> PTHREAD_MUTEX_STALLED_NP);
> pthread_mutexattr_settype (&attr, PTHREAD_MUTEX_ERRORCHECK);
> if (pthread_mutex_init (pmutex, &attr) != 0) {
> printf("Init mutex failed, err=%s\n", strerror(errno));
> pthread_mutexattr_destroy (&attr);
> return NULL;
> }
> }
> #endif
> return pmutex;
> }
>
> void long_running_task(int flag)
> {
> static int counter = 0;
> if (flag == PARENT)
> usleep(5*sleep_period);
> else
> usleep(3*sleep_period);
> counter = (counter + 1) % 100;
> printf("%d: completed %d computing\n", getpid(), counter);
> }
>
> void sig_handler(int signum)
> {
> close(shmid);
> shm_unlink(shm_name);
> exit(0);
> }
>
> int main(int argc, char *argv[])
> {
> pthread_mutex_t *mutex_parent, *mutex_child;
> // signal(SIGUSR1, sig_handler);
> // if (fork()) {
> /* parent process */
> if ((mutex_parent = shmem_mutex_init(SHM_INIT)) == NULL) {
> printf("failed to get the shmem_mutex\n");
> exit(-1);
> }
> #if 0
> while (1) {
> printf("%d: try to hold the lock\n", getpid());
> pthread_mutex_lock(mutex_parent);
> printf("%d: got the lock\n", getpid());
> long_running_task(PARENT);
> pthread_mutex_unlock(mutex_parent);
> printf("%d: released the lock\n", getpid());
> }
> #endif
> // } else {
> #if 0
> /* child process */
> usleep(sleep_period);
> if ((mutex_child = shmem_mutex_init(SHM_GET)) == NULL) {
> printf("failed to get the shmem_mutex\n");
> exit(-1);
> }
> while (1) {
> printf("%d: try to hold the lock\n", getpid());
> pthread_mutex_lock(mutex_child);
> printf("%d: got the lock\n", getpid());
> long_running_task(CHILD);
> pthread_mutex_unlock(mutex_child);
> printf("%d: released the lock\n", getpid());
> }
> #endif
> // }
> return 0;
> }
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Crash when memset of shared mapped memory in ARM
2011-11-16 5:35 Crash when memset of shared mapped memory in ARM naveen yadav
2011-11-16 8:27 ` Russell King - ARM Linux
@ 2011-11-16 8:45 ` Russell King - ARM Linux
2011-11-16 11:54 ` Will Deacon
1 sibling, 1 reply; 6+ messages in thread
From: Russell King - ARM Linux @ 2011-11-16 8:45 UTC (permalink / raw)
To: naveen yadav; +Cc: linux-mm, linux-arm-kernel, linux-kernel
Please do not spam mailing lists -request addresses when you post. The
-request addresses are there for you to give the mailing list software
_instructions_ on what to do with your subscription. It is not for
posts to the mailing list.
Thanks.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Crash when memset of shared mapped memory in ARM
2011-11-16 8:27 ` Russell King - ARM Linux
@ 2011-11-16 9:07 ` naveen yadav
0 siblings, 0 replies; 6+ messages in thread
From: naveen yadav @ 2011-11-16 9:07 UTC (permalink / raw)
To: Russell King - ARM Linux
Cc: linux-mm, linux-arm-kernel-request, linux-arm-kernel,
linux-arm-request, linux-kernel
Ok,
On X86, MIPS
#./tc1
mmap: addr 0x20000000
#
#
On ARM (2.6.35.11 ~ 3.0.2)
#tc1
map: addr 0x20000000
tc1: unhandled page fault (7) at 0x20001000, code 0x817
Bus error (core dumped)
#
#
Debug Information
Pid: 179, comm: tc1
CPU: 0 Not tainted (2.6.35.11 #1)
PC is at 0x400da410
LR is at 0x8858
pc : [<400da410>] lr : [<00008858>] psr: 20000010
sp : bed84814 ip : 20001080 fp : 00000000
r10: 40025000 r9 : 00000000 r8 : 00000000
r7 : 20000000 r6 : 20001000 r5 : 00000000 r4 : 0000ef80
r3 : 00000000 r2 : 20010000 r1 : 00000000 r0 : 20000000
Flags: nzCv IRQs on FIQs on Mode USER_32 ISA ARM Segment user
Process Memory map
00008000-00009000 r-xp 00000000 08:01 153 /usb/sda1/tc1
00010000-00011000 rw-p 00000000 08:01 153 /usb/sda1/tc1
00011000-00032000 rw-p 00011000 08:01 153
20000000-20010000 rw-s 00000000 00:0d 249 /dev/shm/shmem_1234
40000000-4001d000 r-xp 00000000 8b:06 163 /lib/ld-2.11.1.so
4001d000-40021000 rw-p 4001d000 8b:06 163
40024000-40025000 r--p 0001c000 8b:06 163 /lib/ld-2.11.1.so
40025000-40026000 rw-p 0001d000 8b:06 163 /lib/ld-2.11.1.so
40026000-4003b000 r-xp 00000000 8b:06 171 /lib/libpthread-2.11.1.so
4003b000-40042000 ---p 00015000 8b:06 171 /lib/libpthread-2.11.1.so
40042000-40043000 r--p 00014000 8b:06 171 /lib/libpthread-2.11.1.so
40043000-40044000 rw-p 00015000 8b:06 171 /lib/libpthread-2.11.1.so
40044000-40046000 rw-p 40044000 8b:06 171
40046000-4004c000 r-xp 00000000 8b:06 161 /lib/librt-2.11.1.so
4004c000-40053000 ---p 00006000 8b:06 161 /lib/librt-2.11.1.so
40053000-40054000 r--p 00005000 8b:06 161 /lib/librt-2.11.1.so
40054000-40055000 rw-p 00006000 8b:06 161 /lib/librt-2.11.1.so
40055000-4005f000 r-xp 00000000 8b:06 165 /lib/libgcc_s.so.1
4005f000-40067000 ---p 0000a000 8b:06 165 /lib/libgcc_s.so.1
40067000-40068000 rw-p 0000a000 8b:06 165 /lib/libgcc_s.so.1
40068000-40185000 r-xp 00000000 8b:06 174 /lib/libc-2.11.1.so
40185000-4018d000 ---p 0011d000 8b:06 174 /lib/libc-2.11.1.so
4018d000-4018f000 r--p 0011d000 8b:06 174 /lib/libc-2.11.1.so
4018f000-40190000 rw-p 0011f000 8b:06 174 /lib/libc-2.11.1.so
40190000-40193000 rw-p 40190000 8b:06 174
bed63000-bed85000 rw-p befde000 8b:06 174
We wish to know why the same application crashes on ARM ?
On Wed, Nov 16, 2011 at 1:57 PM, Russell King - ARM Linux
<linux@arm.linux.org.uk> wrote:
> On Wed, Nov 16, 2011 at 11:05:49AM +0530, naveen yadav wrote:
>> Hi All,
>>
>> I am running below Test program on ARM cortex a9/a8 on kernel version
>> 2.6.35.14 as well as on 3.0.
>>
>> Please find the test case where:
>>
>> 1. Create shared memory object using shm_open(If we use normal open
>> then no problem only problem with shm_open)
>>
>> 2. ftruncate to given size
>>
>> 3. memory map the shared object to given memory address ( I haved
>> tested without MAP_SHARED, MAP_FIXED as well, problem exist)
>>
>> 4. Memset the shared memory (got page fault when accessing the second page)
>>
>>
>>
>>
>> Observation: Only observed in ARM ( i.e not present in MIPS and X86)
>
> Yes, so, what happens?
>
>>
>>
>> #undef NDEBUG
>> #define _GNU_SOURCE
>> #include <unistd.h>
>> #include <stdio.h>
>> #include <stdlib.h>
>> #include <sys/ipc.h>
>> #include <sys/shm.h>
>> #include <errno.h>
>> #include <pthread.h>
>> #include <string.h>
>> #include <signal.h>
>> #include <fcntl.h>
>> #include <sys/types.h>
>> #include <sys/mman.h>
>>
>> enum {
>> SHM_INIT,
>> SHM_GET
>> };
>>
>> enum {
>> PARENT,
>> CHILD
>> };
>>
>> #define FIXED_MMAP_ADDR 0x20000000
>> #define MMAP_SIZE 0x10000
>>
>> static int shmid;
>> static char shm_name[100];
>> static int sleep_period = 100000;
>> void * shmem_init(int flag)
>> {
>> int start = FIXED_MMAP_ADDR;
>> int memory_size = MMAP_SIZE;
>> int mode = 0666;
>> void *addr;
>> int ret;
>> sprintf(shm_name, "/shmem_1234");
>> shmid = shm_open (shm_name, O_RDWR | O_EXCL | O_CREAT | O_TRUNC, mode);
>> if (shmid < 0) {
>> if (errno == EEXIST) {
>> printf ("shm_open: %s\n", strerror(errno));
>> shmid = shm_open (shm_name, O_RDWR, mode);
>>
>> } else {
>> printf("failed to shm_open, err=%s\n", strerror(errno));
>> return NULL;
>> }
>> }
>> ret = fcntl (shmid, F_SETFD, FD_CLOEXEC);
>> if (ret < 0) {
>> printf("fcntl: %s\n", strerror(errno));
>> return NULL;
>> }
>> ret = ftruncate (shmid, memory_size);
>> if (ret < 0) {
>> printf("ftruncate: %s\n", strerror(errno));
>> return NULL;
>> }
>> addr = mmap ((void *)start, memory_size, PROT_READ | PROT_WRITE,
>> MAP_SHARED | MAP_FIXED, shmid, 0);
>> if (addr == MAP_FAILED) {
>> printf ("mmap: %s\n", strerror(errno));
>> close (shmid);
>> shm_unlink (shm_name);
>> return NULL;
>> }
>>
>> if (flag == SHM_INIT){
>> printf ("mmap: addr %p\n", addr);
>> /* memset on arm creates a unhandled page fault, works fine on mips */
>> memset(addr, 0, memory_size);
>> }
>> return (void *)addr;
>> }
>>
>> pthread_mutex_t * shmem_mutex_init(int flag)
>> {
>> pthread_mutex_t * pmutex = (pthread_mutex_t *)shmem_init(flag);
>> #if 0
>> pthread_mutexattr_t attr;
>> if (flag == SHM_INIT) {
>> pthread_mutexattr_init (&attr);
>> pthread_mutexattr_setpshared (&attr, PTHREAD_PROCESS_SHARED);
>> pthread_mutexattr_setprotocol (&attr, PTHREAD_PRIO_INHERIT);
>> pthread_mutexattr_setrobust_np (&attr,
>> PTHREAD_MUTEX_STALLED_NP);
>> pthread_mutexattr_settype (&attr, PTHREAD_MUTEX_ERRORCHECK);
>> if (pthread_mutex_init (pmutex, &attr) != 0) {
>> printf("Init mutex failed, err=%s\n", strerror(errno));
>> pthread_mutexattr_destroy (&attr);
>> return NULL;
>> }
>> }
>> #endif
>> return pmutex;
>> }
>>
>> void long_running_task(int flag)
>> {
>> static int counter = 0;
>> if (flag == PARENT)
>> usleep(5*sleep_period);
>> else
>> usleep(3*sleep_period);
>> counter = (counter + 1) % 100;
>> printf("%d: completed %d computing\n", getpid(), counter);
>> }
>>
>> void sig_handler(int signum)
>> {
>> close(shmid);
>> shm_unlink(shm_name);
>> exit(0);
>> }
>>
>> int main(int argc, char *argv[])
>> {
>> pthread_mutex_t *mutex_parent, *mutex_child;
>> // signal(SIGUSR1, sig_handler);
>> // if (fork()) {
>> /* parent process */
>> if ((mutex_parent = shmem_mutex_init(SHM_INIT)) == NULL) {
>> printf("failed to get the shmem_mutex\n");
>> exit(-1);
>> }
>> #if 0
>> while (1) {
>> printf("%d: try to hold the lock\n", getpid());
>> pthread_mutex_lock(mutex_parent);
>> printf("%d: got the lock\n", getpid());
>> long_running_task(PARENT);
>> pthread_mutex_unlock(mutex_parent);
>> printf("%d: released the lock\n", getpid());
>> }
>> #endif
>> // } else {
>> #if 0
>> /* child process */
>> usleep(sleep_period);
>> if ((mutex_child = shmem_mutex_init(SHM_GET)) == NULL) {
>> printf("failed to get the shmem_mutex\n");
>> exit(-1);
>> }
>> while (1) {
>> printf("%d: try to hold the lock\n", getpid());
>> pthread_mutex_lock(mutex_child);
>> printf("%d: got the lock\n", getpid());
>> long_running_task(CHILD);
>> pthread_mutex_unlock(mutex_child);
>> printf("%d: released the lock\n", getpid());
>> }
>> #endif
>> // }
>> return 0;
>> }
>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Crash when memset of shared mapped memory in ARM
2011-11-16 8:45 ` Russell King - ARM Linux
@ 2011-11-16 11:54 ` Will Deacon
2011-11-17 12:10 ` Dave Martin
0 siblings, 1 reply; 6+ messages in thread
From: Will Deacon @ 2011-11-16 11:54 UTC (permalink / raw)
To: Russell King - ARM Linux
Cc: naveen yadav, linux-mm, linux-kernel@vger.kernel.org,
linux-arm-kernel@lists.infradead.org
(Replying to Russell so as to lose the -request addresses)
On Wed, Nov 16, 2011 at 08:45:47AM +0000, Russell King - ARM Linux wrote:
> Please do not spam mailing lists -request addresses when you post. The
> -request addresses are there for you to give the mailing list software
> _instructions_ on what to do with your subscription. It is not for
> posts to the mailing list.
For what it's worth, I was brave/daft enough to compile and run the testcase
with an -rc1 kernel and Linaro 11.09 filesystem on the quad A9 Versatile
Express:
root@dancing-fool:~# ./yad
mmap: addr 0x20000000
root@dancing-fool:~# ./yad
shm_open: File exists
mmap: addr 0x20000000
Looks like I'm missing the fireworks, despite the weirdy MAP_SHARED |
MAP_FIXED mmap flags.
Will
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Crash when memset of shared mapped memory in ARM
2011-11-16 11:54 ` Will Deacon
@ 2011-11-17 12:10 ` Dave Martin
0 siblings, 0 replies; 6+ messages in thread
From: Dave Martin @ 2011-11-17 12:10 UTC (permalink / raw)
To: Will Deacon
Cc: Russell King - ARM Linux, linux-arm-kernel@lists.infradead.org,
linux-mm, linux-kernel@vger.kernel.org, naveen yadav
On Wed, Nov 16, 2011 at 11:54:35AM +0000, Will Deacon wrote:
> (Replying to Russell so as to lose the -request addresses)
>
> On Wed, Nov 16, 2011 at 08:45:47AM +0000, Russell King - ARM Linux wrote:
> > Please do not spam mailing lists -request addresses when you post. The
> > -request addresses are there for you to give the mailing list software
> > _instructions_ on what to do with your subscription. It is not for
> > posts to the mailing list.
>
> For what it's worth, I was brave/daft enough to compile and run the testcase
> with an -rc1 kernel and Linaro 11.09 filesystem on the quad A9 Versatile
> Express:
>
> root@dancing-fool:~# ./yad
> mmap: addr 0x20000000
> root@dancing-fool:~# ./yad
> shm_open: File exists
> mmap: addr 0x20000000
>
> Looks like I'm missing the fireworks, despite the weirdy MAP_SHARED |
> MAP_FIXED mmap flags.
I did't see any problem on a random 3.1-rc4 kernel either on A9...
(Since when was MAP_SHARED "weirdy"...?)
Cheers
---Dave
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2011-11-17 12:10 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-11-16 5:35 Crash when memset of shared mapped memory in ARM naveen yadav
2011-11-16 8:27 ` Russell King - ARM Linux
2011-11-16 9:07 ` naveen yadav
2011-11-16 8:45 ` Russell King - ARM Linux
2011-11-16 11:54 ` Will Deacon
2011-11-17 12:10 ` Dave Martin
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).