> On Nov 14, 2021, at 4:58 PM, Trond Myklebust <trondmy@xxxxxxxxxxxxxxx> wrote:
>
> On Sun, 2021-11-14 at 15:16 -0500, Chuck Lever wrote:
>> rtm@xxxxxxxxxxxxx reports:
>>> nfsd4_decode_bitmap4() will write beyond bmval[bmlen-1] if the RPC
>>> directs it to do so. This can cause nfsd4_decode_state_protect4_a()
>>> to write client-supplied data beyond the end of
>>> nfsd4_exchange_id.spo_must_allow[] when called by
>>> nfsd4_decode_exchange_id().
>>
>> Rewrite the loops so nfsd4_decode_bitmap() cannot iterate beyond
>> @bmlen.
>>
>> Reported by: rtm@xxxxxxxxxxxxx
>> Fixes: d1c263a031e8 ("NFSD: Replace READ* macros in
>> nfsd4_decode_fattr()")
>> Signed-off-by: Chuck Lever <chuck.lever@xxxxxxxxxx>
>> ---
>> fs/nfsd/nfs4xdr.c | 7 ++-----
>> 1 file changed, 2 insertions(+), 5 deletions(-)
>>
>> Hi Bruce-
>>
>> This version is cleaned up slightly and has been tested as follows:
>>
>> - I am not able to crash a patched server using rtm's nfsd_1
>> - No new FAILUREs with NFSv4.1 pynfs tests
>> - No new failures with xfstests on NFSv4.1 or NFSv4.2
>> - No new failures with the git regression suite on NFSv4.1 or NFSv4.2
>> - No reports of NFS4ERR_BADXDR or GARBAGE_ARGS during xfs or git regr
>>
>> Hopefully that exercises enough uses of nfsd4_decode_bitmap() to be
>> confident that it is working properly.
>>
>> I tested this on top of my XDR tracepoint series, so the line numbers
>> might be a little off from your current tree. However, it should
>> otherwise apply cleanly.
>>
>> diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
>> index 5ff481e9c85d..479d3452ce6c 100644
>> --- a/fs/nfsd/nfs4xdr.c
>> +++ b/fs/nfsd/nfs4xdr.c
>> @@ -288,11 +288,8 @@ nfsd4_decode_bitmap4(struct nfsd4_compoundargs
>> *argp, u32 *bmval, u32 bmlen)
>> p = xdr_inline_decode(argp->xdr, count << 2);
>> if (!p)
>> return nfserr_bad_xdr;
>> - i = 0;
>> - while (i < count)
>> - bmval[i++] = be32_to_cpup(p++);
>> - while (i < bmlen)
>> - bmval[i++] = 0;
>> + for (i = 0; i < bmlen; i++)
>> + bmval[i] = (i < count) ? be32_to_cpup(p++) : 0;
>>
>> return nfs_ok;
>> }
>>
>
> Why not just convert it to use xdr_stream_decode_uint32_array() instead
> of maintaining this separate open coded version?
Pulling the chain a little more, perhaps all decode_bitmap4 call
sites should be converted to use xdr_stream_decode_uint32_array.
We could consider that as a clean up later; the current fix has
been kept surgical so it can be backported.
--
Chuck Lever
