[PATCH V2] Fix NULL dereference in super_by_fd

[PATCH V2] Fix NULL dereference in super_by_fd

[lixiaokeng]

When we create 100 partitions(major is 259 not 254) in a raid device,
mdadm may coredump:

Core was generated by `/usr/sbin/mdadm --detail --export /dev/md1p3'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0  __strlen_sse2 ()
    at ../sysdeps/x86_64/multiarch/strlen-vec.S:126
126		movdqu	(%rax), %xmm4
(gdb) bt
#0  __strlen_sse2 ()
    at ../sysdeps/x86_64/multiarch/strlen-vec.S:126
#1  0x00007f1944659139 in __strcpy_chk (
    dest=dest@entry=0x55ea8d7c23ac "", src=0x0,
    destlen=destlen@entry=32) at strcpy_chk.c:28
#2  0x000055ea8d10b66d in strcpy (__src=<optimized out>,
    __dest=0x55ea8d7c23ac "")
    at /usr/include/bits/string_fortified.h:79
#3  super_by_fd (fd=fd@entry=3,
    subarrayp=subarrayp@entry=0x7ffe6a1dff08) at util.c:1289
#4  0x000055ea8d11b3a6 in Detail (
    dev=0x7ffe6a1e2f22 "/dev/md1p3", c=0x7ffe6a1e1700)
    at Detail.c:101
#5  0x000055ea8d101e61 in misc_list (c=<optimized out>,
    ss=<optimized out>, dump_directory=<optimized out>,
    ident=<optimized out>, devlist=<optimized out>)
    at mdadm.c:1959
#6  main (argc=<optimized out>, argv=<optimized out>)
    at mdadm.c:1629

The direct cause is fd2devnm return NULL. Here add a check.

Signed-off-by:Lixiaokeng<lixiaokeng@huawei.com>
Signed-off-by:Wuguanghao<wuguanghao3@huawei.com>
---
 mapfile.c | 4 ++++
 util.c    | 7 ++++++-
 2 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/mapfile.c b/mapfile.c
index 8d7acb3..f72fe0d 100644
--- a/mapfile.c
+++ b/mapfile.c
@@ -292,6 +292,10 @@ struct map_ent *map_by_uuid(struct map_ent **map, int uuid[4])
 struct map_ent *map_by_devnm(struct map_ent **map, char *devnm)
 {
 	struct map_ent *mp;
+
+	if (!devnm)
+		return NULL;
+
 	if (!*map)
 		map_read(map);

diff --git a/util.c b/util.c
index 64dd409..3a84ee3 100644
--- a/util.c
+++ b/util.c
@@ -1241,6 +1241,11 @@ struct supertype *super_by_fd(int fd, char **subarrayp)
 	int i;
 	char *subarray = NULL;
 	char container[32] = "";
+	char *devnm = NULL;
+
+	devnm = fd2devnm(fd);
+	if (!devnm)
+		return NULL;

 	sra = sysfs_read(fd, NULL, GET_VERSION);

@@ -1286,7 +1291,7 @@ struct supertype *super_by_fd(int fd, char **subarrayp)
 		if (subarrayp)
 			*subarrayp = subarray;
 		strcpy(st->container_devnm, container);
-		strcpy(st->devnm, fd2devnm(fd));
+		strcpy(st->devnm, devnm);
 	} else
 		free(subarray);

-- 
1.8.3.1

Re: [PATCH V2] Fix NULL dereference in super_by_fd

[Paul Menzel]

Dear lixiaokeng,


Thank you for your patch. I found a few style issues.

Am 14.12.22 um 04:17 schrieb lixiaokeng:
> When we create 100 partitions(major is 259 not 254) in a raid device,

Please add a space before the (.

> mdadm may coredump:
> 
> Core was generated by `/usr/sbin/mdadm --detail --export /dev/md1p3'.
> Program terminated with signal SIGSEGV, Segmentation fault.
> #0  __strlen_sse2 ()
>      at ../sysdeps/x86_64/multiarch/strlen-vec.S:126
> 126		movdqu	(%rax), %xmm4
> (gdb) bt
> #0  __strlen_sse2 ()
>      at ../sysdeps/x86_64/multiarch/strlen-vec.S:126
> #1  0x00007f1944659139 in __strcpy_chk (
>      dest=dest@entry=0x55ea8d7c23ac "", src=0x0,
>      destlen=destlen@entry=32) at strcpy_chk.c:28
> #2  0x000055ea8d10b66d in strcpy (__src=<optimized out>,
>      __dest=0x55ea8d7c23ac "")
>      at /usr/include/bits/string_fortified.h:79
> #3  super_by_fd (fd=fd@entry=3,
>      subarrayp=subarrayp@entry=0x7ffe6a1dff08) at util.c:1289
> #4  0x000055ea8d11b3a6 in Detail (
>      dev=0x7ffe6a1e2f22 "/dev/md1p3", c=0x7ffe6a1e1700)
>      at Detail.c:101
> #5  0x000055ea8d101e61 in misc_list (c=<optimized out>,
>      ss=<optimized out>, dump_directory=<optimized out>,
>      ident=<optimized out>, devlist=<optimized out>)
>      at mdadm.c:1959
> #6  main (argc=<optimized out>, argv=<optimized out>)
>      at mdadm.c:1629

Please do not wrap the pasted lines.

> The direct cause is fd2devnm return NULL. Here add a check.

… returning NULL, so add a check.

> Signed-off-by:Lixiaokeng<lixiaokeng@huawei.com>
> Signed-off-by:Wuguanghao<wuguanghao3@huawei.com>

Please add a space after the colon, and before the <. Also, is 
Lixiaokeng your name, or could it be written Li Xiao Keng? Please use 
that format – `git config --global user.name "…"`.

     git commit --amend --author="Li Xiao Keng <lixiaokeng@huawei.com>"

> ---

As this is the second iteration (version 2) of the patch, it’s common, 
that you add a note, what the difference between this and the previous 
versions are.

>   mapfile.c | 4 ++++
>   util.c    | 7 ++++++-
>   2 files changed, 10 insertions(+), 1 deletion(-)

[…]


Kind regards,

Paul

Re: [PATCH V2] Fix NULL dereference in super_by_fd

[lixiaokeng]

On 2022/12/14 15:55, Paul Menzel wrote:
> Dear lixiaokeng,
> 
> 
> Thank you for your patch. I found a few style issues.
> 
> Am 14.12.22 um 04:17 schrieb lixiaokeng:
>> When we create 100 partitions(major is 259 not 254) in a raid device,
> 
> Please add a space before the (.
> 
>> mdadm may coredump:
>>
>> Core was generated by `/usr/sbin/mdadm --detail --export /dev/md1p3'.
>> Program terminated with signal SIGSEGV, Segmentation fault.
>> #0  __strlen_sse2 ()
>>      at ../sysdeps/x86_64/multiarch/strlen-vec.S:126
>> 126        movdqu    (%rax), %xmm4
>> (gdb) bt
>> #0  __strlen_sse2 ()
>>      at ../sysdeps/x86_64/multiarch/strlen-vec.S:126
>> #1  0x00007f1944659139 in __strcpy_chk (
>>      dest=dest@entry=0x55ea8d7c23ac "", src=0x0,
>>      destlen=destlen@entry=32) at strcpy_chk.c:28
>> #2  0x000055ea8d10b66d in strcpy (__src=<optimized out>,
>>      __dest=0x55ea8d7c23ac "")
>>      at /usr/include/bits/string_fortified.h:79
>> #3  super_by_fd (fd=fd@entry=3,
>>      subarrayp=subarrayp@entry=0x7ffe6a1dff08) at util.c:1289
>> #4  0x000055ea8d11b3a6 in Detail (
>>      dev=0x7ffe6a1e2f22 "/dev/md1p3", c=0x7ffe6a1e1700)
>>      at Detail.c:101
>> #5  0x000055ea8d101e61 in misc_list (c=<optimized out>,
>>      ss=<optimized out>, dump_directory=<optimized out>,
>>      ident=<optimized out>, devlist=<optimized out>)
>>      at mdadm.c:1959
>> #6  main (argc=<optimized out>, argv=<optimized out>)
>>      at mdadm.c:1629
> 
> Please do not wrap the pasted lines.
> 
>> The direct cause is fd2devnm return NULL. Here add a check.
> 
> … returning NULL, so add a check.
> 
>> Signed-off-by:Lixiaokeng<lixiaokeng@huawei.com>
>> Signed-off-by:Wuguanghao<wuguanghao3@huawei.com>
> 
> Please add a space after the colon, and before the <. Also, is Lixiaokeng your name, or could it be written Li Xiao Keng? Please use that format – `git config --global user.name "…"`.
> 
>     git commit --amend --author="Li Xiao Keng <lixiaokeng@huawei.com>"
> 
>> ---
> 
> As this is the second iteration (version 2) of the patch, it’s common, that you add a note, what the difference between this and the previous versions are.
> 
>>   mapfile.c | 4 ++++
>>   util.c    | 7 ++++++-
>>   2 files changed, 10 insertions(+), 1 deletion(-)
> 
> […]
> 
> 
> Kind regards,
> 

Thanks for your advices. I will change it and send it V3 again.

> Paul
> 
> .

Re: [PATCH V2] Fix NULL dereference in super_by_fd

[Mariusz Tkaczyk]

On Wed, 14 Dec 2022 11:17:41 +0800
lixiaokeng <lixiaokeng@huawei.com> wrote:

> strcpy(st->devnm, devnm);

Hi,
Please use strncpy or snprintf here.

Thanks,
Mariusz

Re: [PATCH V2] Fix NULL dereference in super_by_fd

[lixiaokeng]

On 2022/12/15 19:50, Mariusz Tkaczyk wrote:
> On Wed, 14 Dec 2022 11:17:41 +0800
> lixiaokeng <lixiaokeng@huawei.com> wrote:
> 
>> strcpy(st->devnm, devnm);
> 
> Hi,
> Please use strncpy or snprintf here.

Thanks for your advice, but the length of devnm is not
a defined value. I will keep it as the old codes.

> 
> Thanks,
> Mariusz
> 
> .
> 

Re: [PATCH V2] Fix NULL dereference in super_by_fd

[Mariusz Tkaczyk]

On Mon, 19 Dec 2022 19:50:52 +0800
lixiaokeng <lixiaokeng@huawei.com> wrote:

> On 2022/12/15 19:50, Mariusz Tkaczyk wrote:
> > On Wed, 14 Dec 2022 11:17:41 +0800
> > lixiaokeng <lixiaokeng@huawei.com> wrote:
> >   
> >> strcpy(st->devnm, devnm);  
> > 
> > Hi,
> > Please use strncpy or snprintf here.  
> 
> Thanks for your advice, but the length of devnm is not
> a defined value. I will keep it as the old codes.

Supertype devnm is a array defined to be 32.
https://git.kernel.org/pub/scm/utils/mdadm/mdadm.git/tree/mdadm.h#n1256

32 should be changed to MD_NAME_MAX - you can use this define.
I traveled fd2devnm and I can see that at the end devid2devnm returns:
static char devnm[32]
https://git.kernel.org/pub/scm/utils/mdadm/mdadm.git/tree/lib.c#n123

For that reason usage of strcpy in this case seems to be safe, unless we change
something deeper. My recommendation comes from general safe development rules-
we know dest buffer size so we can esnure that it will be ended properly by
'\0', whatever comes to write from fd2devnm().

Thanks,
Mariusz

Re: [PATCH V2] Fix NULL dereference in super_by_fd

[Jes Sorensen]

On 12/19/22 08:08, Mariusz Tkaczyk wrote:
> On Mon, 19 Dec 2022 19:50:52 +0800
> lixiaokeng <lixiaokeng@huawei.com> wrote:
> 
>> On 2022/12/15 19:50, Mariusz Tkaczyk wrote:
>>> On Wed, 14 Dec 2022 11:17:41 +0800
>>> lixiaokeng <lixiaokeng@huawei.com> wrote:
>>>   
>>>> strcpy(st->devnm, devnm);  
>>>
>>> Hi,
>>> Please use strncpy or snprintf here.  
>>
>> Thanks for your advice, but the length of devnm is not
>> a defined value. I will keep it as the old codes.
> 
> Supertype devnm is a array defined to be 32.
> https://git.kernel.org/pub/scm/utils/mdadm/mdadm.git/tree/mdadm.h#n1256
> 
> 32 should be changed to MD_NAME_MAX - you can use this define.
> I traveled fd2devnm and I can see that at the end devid2devnm returns:
> static char devnm[32]
> https://git.kernel.org/pub/scm/utils/mdadm/mdadm.git/tree/lib.c#n123
> 
> For that reason usage of strcpy in this case seems to be safe, unless we change
> something deeper. My recommendation comes from general safe development rules-
> we know dest buffer size so we can esnure that it will be ended properly by
> '\0', whatever comes to write from fd2devnm().

Totally agree here!

Jes

Re: [PATCH V2] Fix NULL dereference in super_by_fd

[lixiaokeng]

On 2022/12/19 21:08, Mariusz Tkaczyk wrote:
> On Mon, 19 Dec 2022 19:50:52 +0800
> lixiaokeng <lixiaokeng@huawei.com> wrote:
> 
>> On 2022/12/15 19:50, Mariusz Tkaczyk wrote:
>>> On Wed, 14 Dec 2022 11:17:41 +0800
>>> lixiaokeng <lixiaokeng@huawei.com> wrote:
>>>   
>>>> strcpy(st->devnm, devnm);  
>>>
>>> Hi,
>>> Please use strncpy or snprintf here.  
>>
>> Thanks for your advice, but the length of devnm is not
>> a defined value. I will keep it as the old codes.
> 
> Supertype devnm is a array defined to be 32.
> https://git.kernel.org/pub/scm/utils/mdadm/mdadm.git/tree/mdadm.h#n1256
> 
> 32 should be changed to MD_NAME_MAX - you can use this define.
> I traveled fd2devnm and I can see that at the end devid2devnm returns:
> static char devnm[32]
> https://git.kernel.org/pub/scm/utils/mdadm/mdadm.git/tree/lib.c#n123
> 
> For that reason usage of strcpy in this case seems to be safe, unless we change
> something deeper. My recommendation comes from general safe development rules-
> we know dest buffer size so we can esnure that it will be ended properly by
> '\0', whatever comes to write from fd2devnm().
> 
> Thanks,
> Mariusz
> 
> 
Thanks, I understand it. I will modify the patch as V4.

Regards,
Li Xiaokeng

> .
>