Re: [PATCH 2/2] hfsplus: bound the wrapper and partition-table hops in hfsplus_read_wrapper()

From: Viacheslav Dubeyko

Date: Mon Sep 28 2026 - 17:03:21 EST


On Sat, 2026-09-26 at 16:40 +0800, Matthias Goergens wrote:
> hfsplus_read_wrapper() rereads the volume header through a bare "goto
> reread", following an HFS wrapper's embedded-volume descriptor or, if
> the header matches neither signature, the partition-table fallback it
> shares with hfs, with no bound on either.  A descriptor or partition
> entry with a zero offset leaves part_start where it was, and the
> mount
> loops forever.
>
> TN1150 ("HFS Plus Volume Format") requires a wrapper's embedded
> volume
> to be HFS Plus: "When an HFS Plus volume is embedded, drEmbedSigWord
> must be kHFSPlusSigWord ('H+')".  So there is at most one wrapper
> hop,
> and Apple's hfs code, GRUB and 7-Zip all follow it once.  A partition
> map sits at the start of the raw device (previous patch), so a
> partition-table hop only makes sense before any wrapper hop.  Allow
> one
> of each, in that order, and fail with -EINVAL otherwise.  A zero
> offset
> is caught on the second pass.
>
> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> Cc: stable@xxxxxxxxxxxxxxx
> Signed-off-by: Matthias Goergens <matthias.goergens@xxxxxxxxx>
> ---
> A wrapper whose embedded-volume descriptor points back at itself
> hangs the mount forever without this patch and fails at once with it:
>
>   img=hfsplus-wrapper-loop.img
>   put() { printf "$2" | dd of=$img bs=1 seek=$1 conv=notrunc
> status=none; }
>   truncate --size=64K $img
>   put $((1024 + 0x00)) '\x42\x44'          # drSigWord 'BD'
>   put $((1024 + 0x0a)) '\x82\x00'          # drAtrb: SLOCK | SPARED
>   put $((1024 + 0x14)) '\x00\x00\x02\x00'  # drAlBlkSiz 512
>   put $((1024 + 0x7c)) '\x48\x2b'          # drEmbedSigWord 'H+'
>   put $((1024 + 0x7e)) '\x00\x00\x00\x64'  # drEmbedExtent: start 0,
> count 100
>   mount -o ro,loop -t hfsplus $img /mnt
>
> The same partition-table entry as the hfs patch hangs an hfsplus
> mount
> the same way:
>
>   img=hfsplus-partmap-loop.img
>   put() { printf "$2" | dd of=$img bs=1 seek=$1 conv=notrunc
> status=none; }
>   truncate --size=64K $img
>   put $((512+0x00)) '\x50\x4d'          # pmSig 'PM'
>   put $((512+0x04)) '\x00\x00\x00\x01'  # pmMapBlkCnt 1
>   put $((512+0x08)) '\x00\x00\x00\x00'  # pmPyPartStart 0 (self)
>   put $((512+0x0c)) '\x00\x00\x00\x64'  # pmPartBlkCnt 100
>   put $((512+0x30)) 'Apple_HFS'         # pmPartType
>   mount -o ro,loop -t hfsplus $img /mnt
> ---
>  fs/hfsplus/wrapper.c | 15 +++++++++++++++
>  1 file changed, 15 insertions(+)
>
> diff --git a/fs/hfsplus/wrapper.c b/fs/hfsplus/wrapper.c
> index 30cf4fe78b3d..df8079a4d8f5 100644
> --- a/fs/hfsplus/wrapper.c
> +++ b/fs/hfsplus/wrapper.c
> @@ -137,6 +137,7 @@ int hfsplus_read_wrapper(struct super_block *sb)
>   struct hfsplus_wd wd;
>   sector_t part_start, part_size;
>   u32 blocksize;
> + bool part_hop_done = false, wrapper_hop_done = false;
>   int error = 0;
>  
>   error = -EINVAL;
> @@ -172,21 +173,35 @@ int hfsplus_read_wrapper(struct super_block
> *sb)
>   case cpu_to_be16(HFSPLUS_VOLHEAD_SIG):
>   break;
>   case cpu_to_be16(HFSP_WRAP_MAGIC):
> + /*
> + * The embedded volume must be HFS Plus (TN1150), so
> a
> + * second wrapper is corrupt; a descriptor pointing
> at
> + * itself would otherwise loop forever.
> + */
> + if (wrapper_hop_done)
> + goto out_free_backup_vhdr;
>   if (!hfsplus_read_mdb(sbi->s_vhdr, &wd))

Ditto. We need to have a proper sanity check in hfsplus_read_mdb().

>   goto out_free_backup_vhdr;
>   wd.ablk_size >>= HFSPLUS_SECTOR_SHIFT;
>   part_start += (sector_t)wd.ablk_start +
>          (sector_t)wd.embed_start *
> wd.ablk_size;
>   part_size = (sector_t)wd.embed_count * wd.ablk_size;
> + wrapper_hop_done = true;
>   goto reread;
>   default:
>   /*
>   * Check for a partition block.
>   *
>   * (should do this only for cdrom/loop though)
> + *
> + * The partition map is at the start of the device:
> follow
> + * it at most once, and not from inside a wrapper.
>   */
> + if (part_hop_done || wrapper_hop_done)
> + goto out_free_backup_vhdr;
>   if (hfs_part_find(sb, &part_start, &part_size))

Ditto. We need to have a proper sanity check in hfs_part_find().

Thanks,
Slava.

>   goto out_free_backup_vhdr;
> + part_hop_done = true;
>   goto reread;
>   }
>