Re: [RFC PATCH 10/17] signed int -> blk64_t to fix bugs at 2^31 - 2^32 blocks

[Valerie Aurora Henson <vaurora@redhat.com>]

On Thu, Nov 13, 2008 at 09:42:52PM -0600, Eric Sandeen wrote:
> Valerie Aurora Henson wrote:
> > On Thu, Nov 13, 2008 at 12:57:42PM -0700, Andreas Dilger wrote:
> >> On Nov 11, 2008  19:43 -0800, Valerie Aurora Henson wrote:
> >>> diff --git a/lib/ext2fs/alloc_tables.c b/lib/ext2fs/alloc_tables.c
> >>> index 7235f7d..71ad445 100644
> >>> --- a/lib/ext2fs/alloc_tables.c
> >>> +++ b/lib/ext2fs/alloc_tables.c
> >>> @@ -147,7 +147,7 @@ errcode_t ext2fs_allocate_group_table(ext2_filsys fs, dgrp_t group,
> >>> -		int prev_block = 0;
> >>> +		blk64_t prev_block = 0;
> >>> @@ -178,7 +178,7 @@ errcode_t ext2fs_allocate_group_table(ext2_filsys fs, dgrp_t group,
> >>>  	if (flexbg_size) {
> >>> -		int prev_block = 0;
> >>> +		blk64_t prev_block = 0;
> >> These appear to be defects in the base code and should be landed ASAP
> >> (as int -> blk_t) independently of this patch series.
> > 
> > Agreed.  Ted, is this a good format for you or do you want me to
> > regenerate against something?
> 
> Is it?
> 
>         if (flexbg_size) {
>                 int prev_block = 0;
>                 if (group && fs->group_desc[group-1].bg_block_bitmap)
>                         prev_block =
> fs->group_desc[group-1].bg_block_bitmap;
>                 start_blk = flexbg_offset(fs, group, prev_block, bmap,
>                                                  0, rem_grps, 1);
>                 last_blk = ext2fs_group_last_block(fs, last_grp);
>         }
> 
> bg_block_bitmap is only a __u32, and that's what we assign to prev_block.

It's a signed/unsigned bug (int vs. blk_t) - it shows up at > 2^31
blocks.

> Just a quick scan, but isn't this just a relative block in the group?

I double-checked and it's relative to the beginning of the file system.

-VAL