* [PATCH] rt: fix shm_open not set ENAMETOOLONG when name exceeds {_POSIX_PATH_MAX}
@ 2023-03-07 9:45 abushwang
2023-03-07 10:02 ` Xi Ruoyao
0 siblings, 1 reply; 6+ messages in thread
From: abushwang @ 2023-03-07 9:45 UTC (permalink / raw)
To: libc-alpha, carlos; +Cc: abushwang
according to man-pages-posix-2017, shm_open() function may fail if the length
of the name argument exceeds {_POSIX_PATH_MAX} and set ENAMETOOLONG
Signed-off-by: abushwang <abushwangs@gmail.com>
---
posix/shm-directory.c | 5 +++--
rt/shm_open.c | 5 +++--
2 files changed, 6 insertions(+), 4 deletions(-)
diff --git a/posix/shm-directory.c b/posix/shm-directory.c
index 86d9fd8e4f..ca9d9f2f77 100644
--- a/posix/shm-directory.c
+++ b/posix/shm-directory.c
@@ -25,6 +25,7 @@
#include <string.h>
#include <sys/mman.h>
#include <fcntl.h>
+#include <errno.h>
int
__shm_get_name (struct shmdir_name *result, const char *name, bool sem_prefix)
@@ -54,9 +55,9 @@ __shm_get_name (struct shmdir_name *result, const char *name, bool sem_prefix)
if (sem_prefix)
alloc_buffer_copy_bytes (&buffer, "sem.", strlen ("sem."));
alloc_buffer_copy_bytes (&buffer, name, namelen + 1);
- if (namelen == 0 || memchr (name, '/', namelen) != NULL
+ if (namelen == 0 || namelen > NAME_MAX || memchr (name, '/', namelen) != NULL
|| alloc_buffer_has_failed (&buffer))
- return -1;
+ return namelen ? ENAMETOOLONG : EINVAL;
return 0;
}
libc_hidden_def (__shm_get_name)
diff --git a/rt/shm_open.c b/rt/shm_open.c
index 6c1f4d604f..fc1dc96bb4 100644
--- a/rt/shm_open.c
+++ b/rt/shm_open.c
@@ -30,9 +30,10 @@ int
__shm_open (const char *name, int oflag, mode_t mode)
{
struct shmdir_name dirname;
- if (__shm_get_name (&dirname, name, false) != 0)
+ int ret =__shm_get_name (&dirname, name, false);
+ if (ret != 0)
{
- __set_errno (EINVAL);
+ __set_errno (ret);
return -1;
}
--
2.36.1
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] rt: fix shm_open not set ENAMETOOLONG when name exceeds {_POSIX_PATH_MAX}
2023-03-07 9:45 [PATCH] rt: fix shm_open not set ENAMETOOLONG when name exceeds {_POSIX_PATH_MAX} abushwang
@ 2023-03-07 10:02 ` Xi Ruoyao
0 siblings, 0 replies; 6+ messages in thread
From: Xi Ruoyao @ 2023-03-07 10:02 UTC (permalink / raw)
To: abushwang, libc-alpha, carlos
On Tue, 2023-03-07 at 17:45 +0800, abushwang via Libc-alpha wrote:
> according to man-pages-posix-2017, shm_open() function may fail if the length
> of the name argument exceeds {_POSIX_PATH_MAX} and set ENAMETOOLONG
>
> Signed-off-by: abushwang <abushwangs@gmail.com>
Please use [PATCH v2] or v3, v4, ... into the title when you update the patch.
/* snip */
> int
> __shm_get_name (struct shmdir_name *result, const char *name, bool sem_prefix)
> @@ -54,9 +55,9 @@ __shm_get_name (struct shmdir_name *result, const char *name, bool sem_prefix)
> if (sem_prefix)
> alloc_buffer_copy_bytes (&buffer, "sem.", strlen ("sem."));
> alloc_buffer_copy_bytes (&buffer, name, namelen + 1);
> - if (namelen == 0 || memchr (name, '/', namelen) != NULL
> + if (namelen == 0 || namelen > NAME_MAX || memchr (name, '/', namelen) != NULL
You can't just reject any namelen > NAMEMAX because a "may" clause in
the specification. You may only do that if the allocation fails.
What you are doing is like: the specification of a capacitor says it may
(not "must" or "shall"!) blow up if the temperature is > 90C, then you
add a bomb into the capacitor which blows up when the temperature
reaches 90C. It's definitely wrong.
I was not telling you to make the code more compact.
--
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH] rt: fix shm_open not set ENAMETOOLONG when name exceeds {_POSIX_PATH_MAX}
@ 2023-03-07 8:32 abushwang
2023-03-07 8:54 ` Xi Ruoyao
0 siblings, 1 reply; 6+ messages in thread
From: abushwang @ 2023-03-07 8:32 UTC (permalink / raw)
To: libc-alpha, carlos; +Cc: abushwang
according to man-pages-posix-2017, shm_open() function may fail if the length
of the name argument exceeds {_POSIX_PATH_MAX} and set ENAMETOOLONG
Signed-off-by: abushwang <abushwangs@gmail.com>
---
posix/shm-directory.c | 5 ++++-
rt/shm_open.c | 5 +++--
2 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/posix/shm-directory.c b/posix/shm-directory.c
index 86d9fd8e4f..93aaeb60a0 100644
--- a/posix/shm-directory.c
+++ b/posix/shm-directory.c
@@ -25,6 +25,7 @@
#include <string.h>
#include <sys/mman.h>
#include <fcntl.h>
+#include <errno.h>
int
__shm_get_name (struct shmdir_name *result, const char *name, bool sem_prefix)
@@ -50,13 +51,15 @@ __shm_get_name (struct shmdir_name *result, const char *name, bool sem_prefix)
while (name[0] == '/')
++name;
namelen = strlen (name);
+ if (namelen > NAME_MAX)
+ return ENAMETOOLONG;
if (sem_prefix)
alloc_buffer_copy_bytes (&buffer, "sem.", strlen ("sem."));
alloc_buffer_copy_bytes (&buffer, name, namelen + 1);
if (namelen == 0 || memchr (name, '/', namelen) != NULL
|| alloc_buffer_has_failed (&buffer))
- return -1;
+ return EINVAL;
return 0;
}
libc_hidden_def (__shm_get_name)
diff --git a/rt/shm_open.c b/rt/shm_open.c
index 6c1f4d604f..fc1dc96bb4 100644
--- a/rt/shm_open.c
+++ b/rt/shm_open.c
@@ -30,9 +30,10 @@ int
__shm_open (const char *name, int oflag, mode_t mode)
{
struct shmdir_name dirname;
- if (__shm_get_name (&dirname, name, false) != 0)
+ int ret =__shm_get_name (&dirname, name, false);
+ if (ret != 0)
{
- __set_errno (EINVAL);
+ __set_errno (ret);
return -1;
}
--
2.36.1
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] rt: fix shm_open not set ENAMETOOLONG when name exceeds {_POSIX_PATH_MAX}
2023-03-07 8:32 abushwang
@ 2023-03-07 8:54 ` Xi Ruoyao
2023-03-07 9:46 ` abush wang
0 siblings, 1 reply; 6+ messages in thread
From: Xi Ruoyao @ 2023-03-07 8:54 UTC (permalink / raw)
To: abushwang, libc-alpha, carlos
On Tue, 2023-03-07 at 16:32 +0800, abushwang via Libc-alpha wrote:
> according to man-pages-posix-2017, shm_open() function may fail if the length
> of the name argument exceeds {_POSIX_PATH_MAX} and set ENAMETOOLONG
It's "may" fail, not "shall" fail. POSIX 2017 says "may" means:
may
Describes a feature or behavior that is optional for an implementation
that conforms to POSIX.1-2017. An application should not rely on the
existence of the feature or behavior. An application that relies on such
a feature or behavior cannot be assured to be portable across conforming
implementations.
We should not break existing programs just for a "may" clause.
/* snip */
> int
> __shm_get_name (struct shmdir_name *result, const char *name, bool
> sem_prefix)
> @@ -50,13 +51,15 @@ __shm_get_name (struct shmdir_name *result, const
> char *name, bool sem_prefix)
> while (name[0] == '/')
> ++name;
> namelen = strlen (name);
> + if (namelen > NAME_MAX)
> + return ENAMETOOLONG;
>
> if (sem_prefix)
> alloc_buffer_copy_bytes (&buffer, "sem.", strlen ("sem."));
> alloc_buffer_copy_bytes (&buffer, name, namelen + 1);
> if (namelen == 0 || memchr (name, '/', namelen) != NULL
> || alloc_buffer_has_failed (&buffer))
If you really want ENAMETOOLONG I guess you can check if namelen >
NAME_MAX here (as a very long namelen would have caused an allocation
failure).
By the way if namelen <= NAME_MAX but alloc_buffer_has_failed (due to a
high memory usage) should we say ENOSPC instead of EINVAL? POSIX 2017
says:
[ENOSPC]
There is insufficient space for the creation of the new shared memory object.
> - return -1;
> + return EINVAL;
> return 0;
> }
--
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] rt: fix shm_open not set ENAMETOOLONG when name exceeds {_POSIX_PATH_MAX}
2023-03-07 8:54 ` Xi Ruoyao
@ 2023-03-07 9:46 ` abush wang
2023-03-07 9:46 ` abush wang
0 siblings, 1 reply; 6+ messages in thread
From: abush wang @ 2023-03-07 9:46 UTC (permalink / raw)
To: Xi Ruoyao; +Cc: libc-alpha, carlos
[-- Attachment #1: Type: text/plain, Size: 2105 bytes --]
I have re-send this patch according to your suggest.
Thinks
abushwang
Xi Ruoyao <xry111@xry111.site> 于2023年3月7日周二 16:54写道：
> On Tue, 2023-03-07 at 16:32 +0800, abushwang via Libc-alpha wrote:
> > according to man-pages-posix-2017, shm_open() function may fail if the
> length
> > of the name argument exceeds {_POSIX_PATH_MAX} and set ENAMETOOLONG
>
> It's "may" fail, not "shall" fail. POSIX 2017 says "may" means:
>
> may
>
> Describes a feature or behavior that is optional for an implementation
> that conforms to POSIX.1-2017. An application should not rely on the
> existence of the feature or behavior. An application that relies on such
> a feature or behavior cannot be assured to be portable across conforming
> implementations.
>
> We should not break existing programs just for a "may" clause.
>
> /* snip */
>
> > int
> > __shm_get_name (struct shmdir_name *result, const char *name, bool
> > sem_prefix)
> > @@ -50,13 +51,15 @@ __shm_get_name (struct shmdir_name *result, const
> > char *name, bool sem_prefix)
> > while (name[0] == '/')
> > ++name;
> > namelen = strlen (name);
> > + if (namelen > NAME_MAX)
> > + return ENAMETOOLONG;
> >
> > if (sem_prefix)
> > alloc_buffer_copy_bytes (&buffer, "sem.", strlen ("sem."));
> > alloc_buffer_copy_bytes (&buffer, name, namelen + 1);
> > if (namelen == 0 || memchr (name, '/', namelen) != NULL
> > || alloc_buffer_has_failed (&buffer))
>
> If you really want ENAMETOOLONG I guess you can check if namelen >
> NAME_MAX here (as a very long namelen would have caused an allocation
> failure).
>
> By the way if namelen <= NAME_MAX but alloc_buffer_has_failed (due to a
> high memory usage) should we say ENOSPC instead of EINVAL? POSIX 2017
> says:
>
> [ENOSPC]
> There is insufficient space for the creation of the new shared memory
> object.
>
> > - return -1;
> > + return EINVAL;
> > return 0;
> > }
>
> --
> Xi Ruoyao <xry111@xry111.site>
> School of Aerospace Science and Technology, Xidian University
>
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH] rt: fix shm_open not set ENAMETOOLONG when name exceeds {_POSIX_PATH_MAX}
2023-03-07 9:46 ` abush wang
@ 2023-03-07 9:46 ` abush wang
0 siblings, 0 replies; 6+ messages in thread
From: abush wang @ 2023-03-07 9:46 UTC (permalink / raw)
To: Xi Ruoyao; +Cc: libc-alpha, carlos
[-- Attachment #1: Type: text/plain, Size: 2389 bytes --]
ENOSPC mean insufficient shared memory object, alloc_buffer_has_failed mean
insufficient memory for name, maybe they are different.
abush wang <abushwangs@gmail.com> 于2023年3月7日周二 17:46写道：
> I have re-send this patch according to your suggest.
>
> Thinks
> abushwang
>
> Xi Ruoyao <xry111@xry111.site> 于2023年3月7日周二 16:54写道：
>
>> On Tue, 2023-03-07 at 16:32 +0800, abushwang via Libc-alpha wrote:
>> > according to man-pages-posix-2017, shm_open() function may fail if the
>> length
>> > of the name argument exceeds {_POSIX_PATH_MAX} and set ENAMETOOLONG
>>
>> It's "may" fail, not "shall" fail. POSIX 2017 says "may" means:
>>
>> may
>>
>> Describes a feature or behavior that is optional for an implementation
>> that conforms to POSIX.1-2017. An application should not rely on the
>> existence of the feature or behavior. An application that relies on such
>> a feature or behavior cannot be assured to be portable across conforming
>> implementations.
>>
>> We should not break existing programs just for a "may" clause.
>>
>> /* snip */
>>
>> > int
>> > __shm_get_name (struct shmdir_name *result, const char *name, bool
>> > sem_prefix)
>> > @@ -50,13 +51,15 @@ __shm_get_name (struct shmdir_name *result, const
>> > char *name, bool sem_prefix)
>> > while (name[0] == '/')
>> > ++name;
>> > namelen = strlen (name);
>> > + if (namelen > NAME_MAX)
>> > + return ENAMETOOLONG;
>> >
>> > if (sem_prefix)
>> > alloc_buffer_copy_bytes (&buffer, "sem.", strlen ("sem."));
>> > alloc_buffer_copy_bytes (&buffer, name, namelen + 1);
>> > if (namelen == 0 || memchr (name, '/', namelen) != NULL
>> > || alloc_buffer_has_failed (&buffer))
>>
>> If you really want ENAMETOOLONG I guess you can check if namelen >
>> NAME_MAX here (as a very long namelen would have caused an allocation
>> failure).
>>
>> By the way if namelen <= NAME_MAX but alloc_buffer_has_failed (due to a
>> high memory usage) should we say ENOSPC instead of EINVAL? POSIX 2017
>> says:
>>
>> [ENOSPC]
>> There is insufficient space for the creation of the new shared memory
>> object.
>>
>> > - return -1;
>> > + return EINVAL;
>> > return 0;
>> > }
>>
>> --
>> Xi Ruoyao <xry111@xry111.site>
>> School of Aerospace Science and Technology, Xidian University
>>
>
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2023-03-07 10:02 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-03-07 9:45 [PATCH] rt: fix shm_open not set ENAMETOOLONG when name exceeds {_POSIX_PATH_MAX} abushwang
2023-03-07 10:02 ` Xi Ruoyao
-- strict thread matches above, loose matches on Subject: below --
2023-03-07 8:32 abushwang
2023-03-07 8:54 ` Xi Ruoyao
2023-03-07 9:46 ` abush wang
2023-03-07 9:46 ` abush wang
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 read-only IMAP folder(s) and NNTP newsgroup(s).