fs/fat.c assumes that 0UL is the same for 32bit and 64bit archs

fs/fat.c assumes that 0UL is the same for 32bit and 64bit archs

[Omniflux]

[-- Attachment #1: Type: text/plain, Size: 287 bytes --]

fs/fat.c assumes that 0UL is the same for 32bit and 64bit archs, but it
is not.

There are four locations in fs/fat.c where a 32bit unsigned int is set
equal to 0UL. This causes warnings when compiling on 64bit architectures
   as 0UL is 0xffffffffffffffff (64bits) there.

-- 
Omniflux

[-- Attachment #2: 0UL-fix.diff --]
[-- Type: text/plain, Size: 1145 bytes --]

--- grub2/fs/fat.c	2004-08-28 07:14:29.000000000 -0600
+++ grub2.modified/fs/fat.c	2005-02-16 00:39:25.271409504 -0700
@@ -243,7 +243,7 @@
   else
     {
       /* FAT12 or FAT16.  */
-      data->root_cluster = ~0UL;
+      data->root_cluster = 0xffffffff;
 
       if (data->num_clusters <= 4085 + 2)
 	{
@@ -293,7 +293,7 @@
 
   /* Start from the root directory.  */
   data->file_cluster = data->root_cluster;
-  data->cur_cluster_num = ~0UL;
+  data->cur_cluster_num = 0xffffffff;
   data->attr = GRUB_FAT_ATTR_DIRECTORY;
   return data;
 
@@ -318,7 +318,7 @@
   
   /* This is a special case. FAT12 and FAT16 doesn't have the root directory
      in clusters.  */
-  if (data->file_cluster == ~0UL)
+  if (data->file_cluster == 0xffffffff)
     {
       size = (data->num_root_sectors << GRUB_DISK_SECTOR_BITS) - offset;
       if (size > len)
@@ -613,7 +613,7 @@
   data->file_size = grub_le_to_cpu32 (dir.file_size);
   data->file_cluster = ((grub_le_to_cpu16 (dir.first_cluster_high) << 16)
 			| grub_le_to_cpu16 (dir.first_cluster_low));
-  data->cur_cluster_num = ~0UL;
+  data->cur_cluster_num = 0xffffffff;
   
   return dirp;
 }

Re: fs/fat.c assumes that 0UL is the same for 32bit and 64bit archs

[Yoshinori K. Okuji]

On Wednesday 16 February 2005 09:56, Omniflux wrote:
> fs/fat.c assumes that 0UL is the same for 32bit and 64bit archs, but
> it is not.
>
> There are four locations in fs/fat.c where a 32bit unsigned int is
> set equal to 0UL. This causes warnings when compiling on 64bit
> architectures as 0UL is 0xffffffffffffffff (64bits) there.

Marco, did you look at this patch?

Okuji

Re: fs/fat.c assumes that 0UL is the same for 32bit and 64bit archs

[Marco Gerards]

"Yoshinori K. Okuji" <okuji@enbug.org> writes:

> On Wednesday 16 February 2005 09:56, Omniflux wrote:
>> fs/fat.c assumes that 0UL is the same for 32bit and 64bit archs, but
>> it is not.
>>
>> There are four locations in fs/fat.c where a 32bit unsigned int is
>> set equal to 0UL. This causes warnings when compiling on 64bit
>> architectures as 0UL is 0xffffffffffffffff (64bits) there.
>
> Marco, did you look at this patch?

Yes.  Sorry, I forgot to reply after I got back. :)

The patch looked fine to me.  I have encountered this problem as well
on 64 bits archs but I ddi not test the patch yet myself, I can't at
the moment.

Because this seems to be the only way to fix this problem, I don't
think there are copyright issues and the patch can be applied.  Right?

Thanks,
Marco

Re: fs/fat.c assumes that 0UL is the same for 32bit and 64bit archs

[Yoshinori K. Okuji]

On Wednesday 02 March 2005 22:20, Marco Gerards wrote:
> Because this seems to be the only way to fix this problem, I don't
> think there are copyright issues and the patch can be applied. 
> Right?

How about this instead of 0xffffffff:

~((grub_uint32_t) 0)

I don't know which is better.

You don't have to care about the copyright, because the change is very 
small.

Okuji