All of lore.kernel.org
 help / color / mirror / Atom feed
* [Xenomai] rt_heap_alloc size 0 - syscall 7 denied
@ 2015-10-14 10:22 Frederik Bayart
  2015-10-14 10:43 ` Philippe Gerum
  0 siblings, 1 reply; 6+ messages in thread
From: Frederik Bayart @ 2015-10-14 10:22 UTC (permalink / raw)
  To: xenomai

Hallo,
I'm using xenomai 3.0 on kernel 3.18.20 (64 bit).

When using single block heap, I'm getting memory allocation problem when calling rt_heap_alloc with size 0. I don't get the error when calling the function with size used in rt_heap_create.
On the mailing list, I found there has been an issue with this in May : http://www.xenomai.org/pipermail/xenomai/2015-May/034165.html
Is this still an issue ?

In attachment a test program htest.c + makefile to generate the problem :

$ sudo ./htest 1  ==> use 0
htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
htest.c:79 ERROR: rt_heap_alloc(-12): Cannot allocate memory

$ sudo ./htest 1  ==> use same size as in create
htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
htest.c:92: heapsize = 70000, usablemem = 70144, usedmem = 70144

However, I both cases, I get a trace in dmesg :

[Xenomai] syscall <7> denied to sysregd

I have no idea why I'm getting this trace. Does anybody has a suggestion ? I found that this error is printed probably in ./kernel/cobalt/posix/syscall.c

If I'm right, syscall 7 would be sc_cobalt_thread_setschedparam_ex (cfr. ./doc/prebuilt/html/xeno3prm/include_2cobalt_2uapi_2syscall_8h_source.html)
I still have other problems which I'm looking at, so maybe they have the same cause.

This is the content of the registry when allocation succeeds :
/var/run/xenomai# find .
.
./root
./root/anon@21808
./root/anon@21808/21808
./root/anon@21808/21808/alchemy
./root/anon@21808/21808/alchemy/heaps
./root/anon@21808/21808/alchemy/heaps/TESTHEAP
./root/anon@21808/system
./root/anon@21808/system/threads
./root/anon@21808/system/heaps
./root/anon@21808/system/version


# xeno-config --info
Xenomai version: Xenomai/cobalt v3.0 -- 
Linux dev1 3.18.20-x86-64-xeno-3.03p.0.0 #1 SMP PREEMPT Tue Oct 13 16:58:44 CEST 2015 x86_64 GNU/Linux
Kernel parameters: BOOT_IMAGE=/vmlinuz-3.18.20-x86-64-xeno-3.03p.0.0 root=UUID=20f2dd2f-f02f-4b82-8538-8fe52e55a8ff ro quiet
I-pipe release #3 detected
Cobalt core 3.0 detected
Compiler: gcc version 4.9.2 (Debian 4.9.2-10) 
Build args: --prefix=/usr --includedir=/usr/include/xenomai --mandir=/usr/share/man --with-testdir=/usr/lib/xenomai/testsuite --with-core=cobalt --enable-smp --enable-pshared --enable-registry --build x86_64-linux-gnu build_alias=x86_64-linux-gnu


I have also attached the kernel config.
Kind regards,
Frederik


-------------- next part --------------
A non-text attachment was scrubbed...
Name: config-3.18.20-x86-64-xeno-3.03p.0.0
Type: application/octet-stream
Size: 128070 bytes
Desc: not available
URL: <http://xenomai.org/pipermail/xenomai/attachments/20151014/7024f00c/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: htest.c
Type: text/x-csrc
Size: 2268 bytes
Desc: not available
URL: <http://xenomai.org/pipermail/xenomai/attachments/20151014/7024f00c/attachment.c>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: makefile
Type: application/octet-stream
Size: 622 bytes
Desc: not available
URL: <http://xenomai.org/pipermail/xenomai/attachments/20151014/7024f00c/attachment-0001.obj>

^ permalink raw reply	[flat|nested] 6+ messages in thread
* Re: [Xenomai] rt_heap_alloc size 0 - syscall 7 denied
  2015-10-14 10:22 [Xenomai] rt_heap_alloc size 0 - syscall 7 denied Frederik Bayart
@ 2015-10-14 10:43 ` Philippe Gerum
  2015-10-14 12:21   ` Frederik Bayart
  0 siblings, 1 reply; 6+ messages in thread
From: Philippe Gerum @ 2015-10-14 10:43 UTC (permalink / raw)
  To: Frederik Bayart, xenomai

On 10/14/2015 12:22 PM, Frederik Bayart wrote:
> Hallo,
> I'm using xenomai 3.0 on kernel 3.18.20 (64 bit).
> 
> When using single block heap, I'm getting memory allocation problem when calling rt_heap_alloc with size 0. I don't get the error when calling the function with size used in rt_heap_create.
> On the mailing list, I found there has been an issue with this in May : http://www.xenomai.org/pipermail/xenomai/2015-May/034165.html
> Is this still an issue ?

No, this one is long gone.

> 
> In attachment a test program htest.c + makefile to generate the problem :
> 
> $ sudo ./htest 1  ==> use 0
> htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
> htest.c:79 ERROR: rt_heap_alloc(-12): Cannot allocate memory
> 
> $ sudo ./htest 1  ==> use same size as in create
> htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
> htest.c:92: heapsize = 70000, usablemem = 70144, usedmem = 70144
>

Does this help?

diff --git a/lib/alchemy/heap.c b/lib/alchemy/heap.c
index 20531e3..146bfa1 100644
--- a/lib/alchemy/heap.c
+++ b/lib/alchemy/heap.c
@@ -442,7 +442,9 @@ int rt_heap_alloc_timed(RT_HEAP *heap,
 		p = __mptr(hcb->sba);
 		if (p)
 			goto done;
-		if (size > 0 && size != hcb->size) {
+		if (size == 0)
+			size = heapobj_size(&hcb->hobj);
+		else if (size != hcb->size) {
 			ret = -EINVAL;
 			goto done;
 		}

> However, I both cases, I get a trace in dmesg :
> 
> [Xenomai] syscall <7> denied to sysregd
> 

That one is a different issue. Does your application have supervisor or
user privileges?

-- 
Philippe.


^ permalink raw reply related	[flat|nested] 6+ messages in thread
* Re: [Xenomai] rt_heap_alloc size 0 - syscall 7 denied
  2015-10-14 10:43 ` Philippe Gerum
@ 2015-10-14 12:21   ` Frederik Bayart
  2015-10-14 12:41     ` Philippe Gerum
  0 siblings, 1 reply; 6+ messages in thread
From: Frederik Bayart @ 2015-10-14 12:21 UTC (permalink / raw)
  To: xenomai

> On Wednesday, 14 October 2015, 12:43, Philippe Gerum <rpm@xenomai.org> wrote:
>On 10/14/2015 12:22 PM, Frederik Bayart wrote:
>> Hallo,
>> I'm using xenomai 3.0 on kernel 3.18.20 (64 bit).
>>
>> When using single block heap, I'm getting memory allocation problem when calling rt_heap_alloc with size 0. I don't get the error when calling the function with size used in rt_heap_create.
>> On the mailing list, I found there has been an issue with this in May : http://www.xenomai.org/pipermail/xenomai/2015-May/034165.html
>> Is this still an issue ?
>
>No, this one is long gone.
>
>>
>> In attachment a test program htest.c + makefile to generate the problem :
>>
>> $ sudo ./htest 1  ==> use 0
>> htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
>> htest.c:79 ERROR: rt_heap_alloc(-12): Cannot allocate memory
>>
>> $ sudo ./htest 1  ==> use same size as in create
>> htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
>> htest.c:92: heapsize = 70000, usablemem = 70144, usedmem = 70144
>>
>
>Does this help?
>
>diff --git a/lib/alchemy/heap.c b/lib/alchemy/heap.c
>index 20531e3..146bfa1 100644
>--- a/lib/alchemy/heap.c
>+++ b/lib/alchemy/heap.c
>@@ -442,7 +442,9 @@ int rt_heap_alloc_timed(RT_HEAP *heap,
>        p = __mptr(hcb->sba);
>        if (p)
>            goto done;
>-        if (size > 0 && size != hcb->size) {
>+        if (size == 0)
>+            size = heapobj_size(&hcb->hobj);
>+        else if (size != hcb->size) {
>            ret = -EINVAL;
>            goto done;
>
>        }
>

After code reading, could it be that the exception for size == 0 has disappeared in the released code ?

  if (hcb->mode & H_SINGLE) {
    p = __mptr(hcb->sba);
    if (p)
      goto done;
    if (size > 0 && size != hcb->size) {
      ret = -EINVAL;
      goto done;
    }
    p = heapobj_alloc(&hcb->hobj, size);
    if (p == NULL) {
      ret = -ENOMEM;
      goto done;
    }
    hcb->sba = __moff(p);
    goto done;
  }


And if size is 0, alloc_block returns NULL so ENOMEM is returned.


>> However, I both cases, I get a trace in dmesg :
>>
>> [Xenomai] syscall <7> denied to sysregd
>
>That one is a different issue. Does your application have supervisor or
>user privileges?
>

I'm running the application as root. Is this the question ?

>-- 
>Philippe.

Frederik

^ permalink raw reply	[flat|nested] 6+ messages in thread
* Re: [Xenomai] rt_heap_alloc size 0 - syscall 7 denied
  2015-10-14 12:21   ` Frederik Bayart
@ 2015-10-14 12:41     ` Philippe Gerum
  2015-10-14 16:55       ` Frederik Bayart
  0 siblings, 1 reply; 6+ messages in thread
From: Philippe Gerum @ 2015-10-14 12:41 UTC (permalink / raw)
  To: Frederik Bayart, xenomai

On 10/14/2015 02:21 PM, Frederik Bayart wrote:
>> On Wednesday, 14 October 2015, 12:43, Philippe Gerum <rpm@xenomai.org> wrote:
>> On 10/14/2015 12:22 PM, Frederik Bayart wrote:
>>> Hallo,
>>> I'm using xenomai 3.0 on kernel 3.18.20 (64 bit).
>>>
>>> When using single block heap, I'm getting memory allocation problem when calling rt_heap_alloc with size 0. I don't get the error when calling the function with size used in rt_heap_create.
>>> On the mailing list, I found there has been an issue with this in May : http://www.xenomai.org/pipermail/xenomai/2015-May/034165.html
>>> Is this still an issue ?
>>
>> No, this one is long gone.
>>
>>>
>>> In attachment a test program htest.c + makefile to generate the problem :
>>>
>>> $ sudo ./htest 1  ==> use 0
>>> htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
>>> htest.c:79 ERROR: rt_heap_alloc(-12): Cannot allocate memory
>>>
>>> $ sudo ./htest 1  ==> use same size as in create
>>> htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
>>> htest.c:92: heapsize = 70000, usablemem = 70144, usedmem = 70144
>>>
>>
>> Does this help?
>>
>> diff --git a/lib/alchemy/heap.c b/lib/alchemy/heap.c
>> index 20531e3..146bfa1 100644
>> --- a/lib/alchemy/heap.c
>> +++ b/lib/alchemy/heap.c
>> @@ -442,7 +442,9 @@ int rt_heap_alloc_timed(RT_HEAP *heap,
>>         p = __mptr(hcb->sba);
>>         if (p)
>>             goto done;
>> -        if (size > 0 && size != hcb->size) {
>> +        if (size == 0)
>> +            size = heapobj_size(&hcb->hobj);
>> +        else if (size != hcb->size) {
>>             ret = -EINVAL;
>>             goto done;
>>
>>         }
>>
> 
> After code reading, could it be that the exception for size == 0 has disappeared in the released code ?
> 

Did you try that patch?

-- 
Philippe.


^ permalink raw reply	[flat|nested] 6+ messages in thread
* Re: [Xenomai] rt_heap_alloc size 0 - syscall 7 denied
  2015-10-14 12:41     ` Philippe Gerum
@ 2015-10-14 16:55       ` Frederik Bayart
  2015-10-15  7:56         ` Philippe Gerum
  0 siblings, 1 reply; 6+ messages in thread
From: Frederik Bayart @ 2015-10-14 16:55 UTC (permalink / raw)
  To: xenomai

>On Wednesday, 14 October 2015, 14:41, Philippe Gerum <rpm@xenomai.org> wrote:
>
>On 10/14/2015 02:21 PM, Frederik Bayart wrote:
>>> On Wednesday, 14 October 2015, 12:43, Philippe Gerum <rpm@xenomai.org> wrote:
>>> On 10/14/2015 12:22 PM, Frederik Bayart wrote:
>>>> Hallo,
>>>> I'm using xenomai 3.0 on kernel 3.18.20 (64 bit).
>>>>
>>>> When using single block heap, I'm getting memory allocation problem when calling rt_heap_alloc with size 0. I don't get the error when calling the function with size used in rt_heap_create.
>>>> On the mailing list, I found there has been an issue with this in May : http://www.xenomai.org/pipermail/xenomai/2015-May/034165.html
>>>> Is this still an issue ?
>>>
>>> No, this one is long gone.
>>>
>>>>
>>>> In attachment a test program htest.c + makefile to generate the problem :
>>>>
>>>> $ sudo ./htest 1  ==> use 0
>>>> htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
>>>> htest.c:79 ERROR: rt_heap_alloc(-12): Cannot allocate memory
>>>>
>>>> $ sudo ./htest 1  ==> use same size as in create
>>>> htest.c:69: heapsize = 70000, usablemem = 70144, usedmem = 0
>>>> htest.c:92: heapsize = 70000, usablemem = 70144, usedmem = 70144
>>>>
>>>
>>> Does this help?
>>>
>>> diff --git a/lib/alchemy/heap.c b/lib/alchemy/heap.c
>>> index 20531e3..146bfa1 100644
>>> --- a/lib/alchemy/heap.c
>>> +++ b/lib/alchemy/heap.c
>>> @@ -442,7 +442,9 @@ int rt_heap_alloc_timed(RT_HEAP *heap,
>>>        p = __mptr(hcb->sba);
>>>        if (p)
>>>            goto done;
>>> -        if (size > 0 && size != hcb->size) {
>>> +        if (size == 0)
>>> +            size = heapobj_size(&hcb->hobj);
>>> +        else if (size != hcb->size) {
>>>            ret = -EINVAL;
>>>            goto done;
>>>
>>>        }
>>>
>>
>> After code reading, could it be that the exception for size == 0 has disappeared in the released code ?
>>
>
>Did you try that patch?

Sorry, I didn't understand that the patch still has to be applied.
So the patch is working, thanks for the quick fix.

However the biggest problem is the denied syscall :

[Xenomai] syscall <7> denied to sysregd[22217]

I'm running the test program as root. Also sysregd is running as root.

root     22292     1  0 16:41 ?        00:00:00 sysregd --daemon --root /var/run/xenomai/root/anon@22289 --anon

So this would be sc_cobalt_thread_setschedparam_ex. So I'm thinking this is called maybe for the main thread shadowing, e.g. for the priority.

If I execute : 
# ./htest –main-prio=15

then I see in /proc/xenomai/sched/threads :

CPU  PID    CLASS  TYPE      PRI   TIMEOUT       STAT       NAME
0  16149  rt     cobalt      0   -             X          htest
0  16152  rt     cobalt      0   -             X          sysregd
0  16154  rt     cobalt      0   -             X          sysregd
0  16156  rt     cobalt      0   -             X          htest


so no prio 15.

The config dump :

# ./htest --dump-config
based on Xenomai/cobalt v3.0 -- 
CONFIG_MMU=1
CONFIG_SMP=1
CONFIG_XENO_BUILD_ARGS=" '--prefix=/usr' '--includedir=/usr/include/xenomai' '--mandir=/usr/share/man' '--with-testdir=/usr/lib/xenomai/testsuite' '--with-core=cobalt' '--enable-smp' '--enable-pshared' '--enable-registry' '--build' 'x86_64-linux-gnu' 'build_alias=x86_64-linux-gnu'"
CONFIG_XENO_BUILD_STRING="x86_64-pc-linux-gnu"
CONFIG_XENO_COBALT=1
CONFIG_XENO_COMPILER="gcc version 4.9.2 (Debian 4.9.2-10) "
CONFIG_XENO_DEFAULT_PERIOD=100000
CONFIG_XENO_FORTIFY=1
CONFIG_XENO_HOST_STRING="x86_64-pc-linux-gnu"
CONFIG_XENO_LORES_CLOCK_DISABLED=1
CONFIG_XENO_PREFIX="/usr"
CONFIG_XENO_PSHARED=1
CONFIG_XENO_RAW_CLOCK_ENABLED=1
CONFIG_XENO_REGISTRY=1
CONFIG_XENO_REGISTRY_ROOT="/var/run/xenomai"
CONFIG_XENO_REVISION_LEVEL=0
CONFIG_XENO_SANITY=1
CONFIG_XENO_TLSF=1
CONFIG_XENO_TLS_MODEL="initial-exec"
CONFIG_XENO_UAPI_LEVEL=14
CONFIG_XENO_VERSION_MAJOR=3
CONFIG_XENO_VERSION_MINOR=0
CONFIG_XENO_VERSION_NAME="Exact Zero"
CONFIG_XENO_VERSION_STRING="3.0"
CONFIG_XENO_X86_VSYSCALL=1
---
CONFIG_XENO_ASYNC_CANCEL is OFF
CONFIG_XENO_COPPERPLATE_CLOCK_RESTRICTED is OFF
CONFIG_XENO_DEBUG is OFF
CONFIG_XENO_DEBUG_FULL is OFF
CONFIG_XENO_LIBS_DLOPEN is OFF
CONFIG_XENO_MERCURY is OFF
CONFIG_XENO_VALGRIND_API is OFF
CONFIG_XENO_WORKAROUND_CONDVAR_PI is OFF
---
PTHREAD_STACK_DEFAULT=65536

Frederik


^ permalink raw reply	[flat|nested] 6+ messages in thread
* Re: [Xenomai] rt_heap_alloc size 0 - syscall 7 denied
  2015-10-14 16:55       ` Frederik Bayart
@ 2015-10-15  7:56         ` Philippe Gerum
  0 siblings, 0 replies; 6+ messages in thread
From: Philippe Gerum @ 2015-10-15  7:56 UTC (permalink / raw)
  To: Frederik Bayart, xenomai

On 10/14/2015 06:55 PM, Frederik Bayart wrote:
> However the biggest problem is the denied syscall :
> 
> [Xenomai] syscall <7> denied to sysregd[22217]
> 

This one is not an issue, it's merely debug noise. You can silence it
with this change; the result will be functionally the same, without the
scary warning.

diff --git a/lib/copperplate/regd/regd.c b/lib/copperplate/regd/regd.c
index 25bd7a5..5d69d83 100644
--- a/lib/copperplate/regd/regd.c
+++ b/lib/copperplate/regd/regd.c
@@ -538,7 +538,7 @@ int main(int argc, char *const *argv)

 	/* Force SCHED_OTHER. */
 	schedp.sched_priority = 0;
-	pthread_setschedparam(pthread_self(), SCHED_OTHER, &schedp);
+	__STD(pthread_setschedparam(pthread_self(), SCHED_OTHER, &schedp));

 	memset(&sa, 0, sizeof(sa));
 	sa.sa_handler = SIG_IGN;

-- 
Philippe.


^ permalink raw reply related	[flat|nested] 6+ messages in thread
end of thread, other threads:[~2015-10-15  7:56 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-10-14 10:22 [Xenomai] rt_heap_alloc size 0 - syscall 7 denied Frederik Bayart
2015-10-14 10:43 ` Philippe Gerum
2015-10-14 12:21   ` Frederik Bayart
2015-10-14 12:41     ` Philippe Gerum
2015-10-14 16:55       ` Frederik Bayart
2015-10-15  7:56         ` Philippe Gerum

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.