Re: clamav 0.96.2

   Hi Axel,

Have been running my own version for the past days without problems.

I have just tried upgrading to your version, but it segmentation faults. 
Perhaps the mailer did something to the patch I sent inline, like 
breaking long lines, so that it does not apply cleanly? I have attatched 
my version of the file.

Regards,
Kim

On 18-08-2010 19:35, Kim Bisgaard wrote:
> I have just seen a new page at clamav:
> https://wiki.clamav.net/Main/UpgradeNotes0962
>
> which has this patch I'm just rebuild and is about to try out - just
> hang on .... :
> --- a/ChangeLog
> +++ b/ChangeLog
> @@ -1,3 +1,7 @@
> +Fri Aug 13 13:57:04 EEST 2010 (edwin)
> +--------------------------------------
> + * libclamav/bytecode_detect.c: prevent fclose(NULL) on SELinux (bb
> #2200)
> +
>  Thu Aug 12 18:49:37 CEST 2010 (tk)
>  ----------------------------------
>   * V 0.96.2
> diff --git a/libclamav/bytecode_detect.c b/libclamav/bytecode_detect.c
> index 6311314..9271fc7 100644
> --- a/libclamav/bytecode_detect.c
> +++ b/libclamav/bytecode_detect.c
> @@ -109,9 +109,11 @@ static int detect_SELinux(void)
>      f = fopen("/selinux/enforce", "r");
>          if (!f && errno == EACCES)
>          return 2;
> -    if (fscanf(f, "%d", &enforce) == 1)
> +    if (f) {
> +        if (fscanf(f, "%d", &enforce) == 1)
>          selinux = 2;
> -    fclose(f);
> +        fclose(f);
> +    }
>      return selinux;
>      }
>      while (fgets(line, sizeof(line), f)) {
> @@ -130,8 +132,8 @@ static int detect_SELinux(void)
>          selinux = 2;
>      if (enforce == -1)
>          selinux = 0;
> +    fclose(f);
>      }
> -    fclose(f);
>      return selinux;
>  }
>
> On 18-08-2010 17:26, Axel Thimm wrote:
>> Hi,
>>
>> On Fri, 2010-08-13 at 11:41 +0200, Kim Bisgaard wrote:
>>> Hi,
>>>
>>> I have tried to build (change the spec to above version) the newly
>>> released clamav-0.96.2. It builds ok once one adds 'BuildRequires:
>>> sendmail-milter', *BUT* here it core dumps once I try to run it.
>>>
>>> I have tried to get the debug-symbols, and starting clamd in gdb but
>>> when started in debugger it does not core dump :-( could be a selinux
>>> thing - but I have not found any AVC's that makes the core dump go
>>> away.
>>>
>>> This is f13 and i686.
>>>
>>> I advice to hold off the update until a patch shows up.
>>
>> Thanks for the heads-up. I did build it, but I placed it under bleeding:
>>
>> http://packages.atrpms.net/name/clamav/
>>
>> Can you check whether this build exhibits the same issues like you
>> describe above? Thanks!
>
> _______________________________________________
> atrpms-devel mailing list
> [email protected]
> http://lists.atrpms.net/mailman/listinfo/atrpms-devel

--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,7 @@
+Fri Aug 13 13:57:04 EEST 2010 (edwin)
+--------------------------------------
+ * libclamav/bytecode_detect.c: prevent fclose(NULL) on SELinux (bb #2200)
+
 Thu Aug 12 18:49:37 CEST 2010 (tk)
 ----------------------------------
  * V 0.96.2
diff --git a/libclamav/bytecode_detect.c b/libclamav/bytecode_detect.c
index 6311314..9271fc7 100644
--- a/libclamav/bytecode_detect.c
+++ b/libclamav/bytecode_detect.c
@@ -109,9 +109,11 @@ static int detect_SELinux(void)
 	f = fopen("/selinux/enforce", "r");
         if (!f && errno == EACCES)
 		return 2;
-	if (fscanf(f, "%d", &enforce) == 1)
+	if (f) {
+	    if (fscanf(f, "%d", &enforce) == 1)
 		selinux = 2;
-	fclose(f);
+	    fclose(f);
+	}
 	return selinux;
     }
     while (fgets(line, sizeof(line), f)) {
@@ -130,8 +132,8 @@ static int detect_SELinux(void)
 	    selinux = 2;
 	if (enforce == -1)
 	    selinux = 0;
+	fclose(f);
     }
-    fclose(f);
     return selinux;
 }