[PATCH] Fix memory access errors for kexec PPC

[PATCH] Fix memory access errors for kexec PPC

[Suzuki Poulose]

Hi,

I was trying the kexec for ppc32 and came across a couple of memory errors
while running with glibc.

The attached patch is the outcome of the glibc's alerts !
We define buf[MAXBYTES-1] and issue  fread(buf, 1, MAXBYTES, file), which glibc
reports an error.

Also there is a typo in the realloc_memory_ranges() code for ppc which causes in
a double free().

Kindly apply.


Thanks
Suzuki


---

Signed-off-by: Suzuki K. Poulose <suzuki@in.ibm.com>

diff --git a/kexec/arch/ppc/crashdump-powerpc.c b/kexec/arch/ppc/crashdump-powerpc.c
index 7bfad20..7853dbe 100644
--- a/kexec/arch/ppc/crashdump-powerpc.c
+++ b/kexec/arch/ppc/crashdump-powerpc.c
@@ -83,7 +83,7 @@ static int get_crash_memory_ranges(struct memory_range **range, int *ranges)
  	int memory_ranges = 0;
  	char device_tree[256] = "/proc/device-tree/";
  	char fname[256];
-	char buf[MAXBYTES-1];
+	char buf[MAXBYTES];
  	DIR *dir, *dmem;
  	FILE *file;
  	struct dirent *dentry, *mentry;
diff --git a/kexec/arch/ppc/kexec-ppc.c b/kexec/arch/ppc/kexec-ppc.c
index ab76d6f..837021f 100644
--- a/kexec/arch/ppc/kexec-ppc.c
+++ b/kexec/arch/ppc/kexec-ppc.c
@@ -196,7 +196,7 @@ static int realloc_memory_ranges(void)
  	if (!memory_range)
  		goto err;
  
-	base_memory_range = (struct memory_range *) realloc(memory_range,
+	base_memory_range = (struct memory_range *) realloc(base_memory_range,
  			memory_range_len);
  	if (!base_memory_range)
  		goto err;
@@ -319,7 +319,7 @@ static int get_devtree_details(unsigned long kexec_flags)
  	unsigned long long htab_base, htab_size;
  	unsigned long long kernel_end;
  	unsigned long long initrd_start, initrd_end;
-	char buf[MAXBYTES-1];
+	char buf[MAXBYTES];
  	char device_tree[256] = "/proc/device-tree/";
  	char fname[256];
  	DIR *dir, *cdir;

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH] Fix memory access errors for kexec PPC

[Simon Horman]

On Wed, Apr 20, 2011 at 03:02:02PM +0530, Suzuki Poulose wrote:
> Hi,
> 
> I was trying the kexec for ppc32 and came across a couple of memory errors
> while running with glibc.
> 
> The attached patch is the outcome of the glibc's alerts !
> We define buf[MAXBYTES-1] and issue  fread(buf, 1, MAXBYTES, file), which glibc
> reports an error.
> 
> Also there is a typo in the realloc_memory_ranges() code for ppc which causes in
> a double free().

Thanks,

could you please make this a formal submission by
supplying a Signed-off-by line?

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH] Fix memory access errors for kexec PPC

[Suzuki Poulose]

On 04/27/11 11:47, Simon Horman wrote:
> On Wed, Apr 20, 2011 at 03:02:02PM +0530, Suzuki Poulose wrote:
>> Hi,
>>
>> I was trying the kexec for ppc32 and came across a couple of memory errors
>> while running with glibc.
>>
>> The attached patch is the outcome of the glibc's alerts !
>> We define buf[MAXBYTES-1] and issue  fread(buf, 1, MAXBYTES, file), which glibc
>> reports an error.
>>
>> Also there is a typo in the realloc_memory_ranges() code for ppc which causes in
>> a double free().
>
> Thanks,
>
> could you please make this a formal submission by
> supplying a Signed-off-by line?
I have updated the patch to use realloc instead of malloc() for memory_range.
---

Fix memory access errors for ppc

The patch fixes memory overflow errors and improper reallocation of memory ranges.

Signed-off-by: Suzuki K. Poulose <suzuki@in.ibm.com>

diff --git a/kexec/arch/ppc/crashdump-powerpc.c b/kexec/arch/ppc/crashdump-powerpc.c
index 7bfad20..7853dbe 100644
--- a/kexec/arch/ppc/crashdump-powerpc.c
+++ b/kexec/arch/ppc/crashdump-powerpc.c
@@ -83,7 +83,7 @@ static int get_crash_memory_ranges(struct memory_range **range, int *ranges)
  	int memory_ranges = 0;
  	char device_tree[256] = "/proc/device-tree/";
  	char fname[256];
-	char buf[MAXBYTES-1];
+	char buf[MAXBYTES];
  	DIR *dir, *dmem;
  	FILE *file;
  	struct dirent *dentry, *mentry;
diff --git a/kexec/arch/ppc/kexec-ppc.c b/kexec/arch/ppc/kexec-ppc.c
index ab76d6f..96fbc12 100644
--- a/kexec/arch/ppc/kexec-ppc.c
+++ b/kexec/arch/ppc/kexec-ppc.c
@@ -192,11 +192,12 @@ static int realloc_memory_ranges(void)
  	max_memory_ranges++;
  	memory_range_len = sizeof(struct memory_range) * max_memory_ranges;
  
-	memory_range = (struct memory_range *) malloc(memory_range_len);
+	memory_range = (struct memory_range *) realloc(memory_range,
+							memory_range_len);
  	if (!memory_range)
  		goto err;
  
-	base_memory_range = (struct memory_range *) realloc(memory_range,
+	base_memory_range = (struct memory_range *) realloc(base_memory_range,
  			memory_range_len);
  	if (!base_memory_range)
  		goto err;
@@ -319,7 +320,7 @@ static int get_devtree_details(unsigned long kexec_flags)
  	unsigned long long htab_base, htab_size;
  	unsigned long long kernel_end;
  	unsigned long long initrd_start, initrd_end;
-	char buf[MAXBYTES-1];
+	char buf[MAXBYTES];
  	char device_tree[256] = "/proc/device-tree/";
  	char fname[256];
  	DIR *dir, *cdir;

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH] Fix memory access errors for kexec PPC

[Simon Horman]

On Wed, Apr 27, 2011 at 12:10:19PM +0530, Suzuki Poulose wrote:
> On 04/27/11 11:47, Simon Horman wrote:
> >On Wed, Apr 20, 2011 at 03:02:02PM +0530, Suzuki Poulose wrote:
> >>Hi,
> >>
> >>I was trying the kexec for ppc32 and came across a couple of memory errors
> >>while running with glibc.
> >>
> >>The attached patch is the outcome of the glibc's alerts !
> >>We define buf[MAXBYTES-1] and issue  fread(buf, 1, MAXBYTES, file), which glibc
> >>reports an error.
> >>
> >>Also there is a typo in the realloc_memory_ranges() code for ppc which causes in
> >>a double free().
> >
> >Thanks,
> >
> >could you please make this a formal submission by
> >supplying a Signed-off-by line?
> I have updated the patch to use realloc instead of malloc() for memory_range.

Hi,

unfortuantely this doesn't seep to apply to the current git tree.
Could you fix that?

The tree is available at
git://git.kernel.org/pub/scm/utils/kernel/kexec/kexec-tools.git

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec