| Message ID | 20210427121903.76556-1-xuanzhuo@linux.alibaba.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | BPF |
| Headers | show |
| Series | [bpf] xsk: fix for xp_aligned_validate_desc() when len == chunk_size | expand |
| Context | Check | Description |
|---|---|---|
| netdev/cover_letter | success | Link |
| netdev/fixes_present | success | Link |
| netdev/patch_count | success | Link |
| netdev/tree_selection | success | Clearly marked for bpf |
| netdev/subject_prefix | success | Link |
| netdev/cc_maintainers | success | CCed 17 of 17 maintainers |
| netdev/source_inline | success | Was 0 now: 0 |
| netdev/verify_signedoff | success | Link |
| netdev/module_param | success | Was 0 now: 0 |
| netdev/build_32bit | success | Errors and warnings before: 1 this patch: 1 |
| netdev/kdoc | success | Errors and warnings before: 0 this patch: 0 |
| netdev/verify_fixes | success | Link |
| netdev/checkpatch | success | total: 0 errors, 0 warnings, 0 checks, 16 lines checked |
| netdev/build_allmodconfig_warn | success | Errors and warnings before: 1 this patch: 1 |
| netdev/header_inline | success | Link |
On Tue, Apr 27, 2021 at 2:19 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote: > > When desc->len is equal to chunk_size, it is legal. But > xp_aligned_validate_desc() got "chunk_end" by desc->addr + desc->len > pointing to the next chunk during the check, which caused the check to > fail. Thanks Xuan for the fix. Off-by-one error. A classic unfortunately. Think your fix also makes it easier to understand the code too, so good. > Fixes: 35fcde7f8deb ("xsk: support for Tx") > Fixes: bbff2f321a86 ("xsk: new descriptor addressing scheme") Just did some quick research and it seems the bug was introduced in the bbff2f321a86 commit above, not the first one 35fcde7f8deb. Or am I mistaken? > Fixes: 2b43470add8c ("xsk: Introduce AF_XDP buffer allocation API") > Fixes: 26062b185eee ("xsk: Explicitly inline functions and move definitions") And in these two, the code was moved around first to a new function in 2b43470add8c, then this function was moved to a new file in 26062b185eee. I believe documenting this in your commit message would make life simpler for the nice people backporting this fix. Or is this implicit in the multiple Fixes tags? Could someone with more experience in these things comment please. Thank you: Magnus Acked-by: Magnus Karlsson <magnus.karlsson@intel.com> > Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com> > --- > net/xdp/xsk_queue.h | 7 +++---- > 1 file changed, 3 insertions(+), 4 deletions(-) > > diff --git a/net/xdp/xsk_queue.h b/net/xdp/xsk_queue.h > index 2823b7c3302d..40f359bf2044 100644 > --- a/net/xdp/xsk_queue.h > +++ b/net/xdp/xsk_queue.h > @@ -128,13 +128,12 @@ static inline bool xskq_cons_read_addr_unchecked(struct xsk_queue *q, u64 *addr) > static inline bool xp_aligned_validate_desc(struct xsk_buff_pool *pool, > struct xdp_desc *desc) > { > - u64 chunk, chunk_end; > + u64 chunk; > > - chunk = xp_aligned_extract_addr(pool, desc->addr); > - chunk_end = xp_aligned_extract_addr(pool, desc->addr + desc->len); > - if (chunk != chunk_end) > + if (desc->len > pool->chunk_size) > return false; > > + chunk = xp_aligned_extract_addr(pool, desc->addr); > if (chunk >= pool->addrs_cnt) > return false; > > -- > 2.31.0 >
On 4/28/21 10:00 AM, Magnus Karlsson wrote: > On Tue, Apr 27, 2021 at 2:19 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote: >> >> When desc->len is equal to chunk_size, it is legal. But >> xp_aligned_validate_desc() got "chunk_end" by desc->addr + desc->len >> pointing to the next chunk during the check, which caused the check to >> fail. > > Thanks Xuan for the fix. Off-by-one error. A classic unfortunately. > > Think your fix also makes it easier to understand the code too, so good. > >> Fixes: 35fcde7f8deb ("xsk: support for Tx") >> Fixes: bbff2f321a86 ("xsk: new descriptor addressing scheme") > > Just did some quick research and it seems the bug was introduced in > the bbff2f321a86 commit above, not the first one 35fcde7f8deb. Or am I > mistaken? > >> Fixes: 2b43470add8c ("xsk: Introduce AF_XDP buffer allocation API") >> Fixes: 26062b185eee ("xsk: Explicitly inline functions and move definitions") > > And in these two, the code was moved around first to a new function in > 2b43470add8c, then this function was moved to a new file in > 26062b185eee. I believe documenting this in your commit message would > make life simpler for the nice people backporting this fix. Or is this > implicit in the multiple Fixes tags? Could someone with more > experience in these things comment please. Fully agree with Magnus, providing more context in the commit message is always appreciated. Xuan, please extend and resubmit. Thanks! > Acked-by: Magnus Karlsson <magnus.karlsson@intel.com> > >> Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com> >> --- >> net/xdp/xsk_queue.h | 7 +++---- >> 1 file changed, 3 insertions(+), 4 deletions(-) >> >> diff --git a/net/xdp/xsk_queue.h b/net/xdp/xsk_queue.h >> index 2823b7c3302d..40f359bf2044 100644 >> --- a/net/xdp/xsk_queue.h >> +++ b/net/xdp/xsk_queue.h >> @@ -128,13 +128,12 @@ static inline bool xskq_cons_read_addr_unchecked(struct xsk_queue *q, u64 *addr) >> static inline bool xp_aligned_validate_desc(struct xsk_buff_pool *pool, >> struct xdp_desc *desc) >> { >> - u64 chunk, chunk_end; >> + u64 chunk; >> >> - chunk = xp_aligned_extract_addr(pool, desc->addr); >> - chunk_end = xp_aligned_extract_addr(pool, desc->addr + desc->len); >> - if (chunk != chunk_end) >> + if (desc->len > pool->chunk_size) >> return false; >> >> + chunk = xp_aligned_extract_addr(pool, desc->addr); >> if (chunk >= pool->addrs_cnt) >> return false; >> >> -- >> 2.31.0 >>
diff --git a/net/xdp/xsk_queue.h b/net/xdp/xsk_queue.h index 2823b7c3302d..40f359bf2044 100644 --- a/net/xdp/xsk_queue.h +++ b/net/xdp/xsk_queue.h @@ -128,13 +128,12 @@ static inline bool xskq_cons_read_addr_unchecked(struct xsk_queue *q, u64 *addr) static inline bool xp_aligned_validate_desc(struct xsk_buff_pool *pool, struct xdp_desc *desc) { - u64 chunk, chunk_end; + u64 chunk; - chunk = xp_aligned_extract_addr(pool, desc->addr); - chunk_end = xp_aligned_extract_addr(pool, desc->addr + desc->len); - if (chunk != chunk_end) + if (desc->len > pool->chunk_size) return false; + chunk = xp_aligned_extract_addr(pool, desc->addr); if (chunk >= pool->addrs_cnt) return false;
When desc->len is equal to chunk_size, it is legal. But xp_aligned_validate_desc() got "chunk_end" by desc->addr + desc->len pointing to the next chunk during the check, which caused the check to fail. Fixes: 35fcde7f8deb ("xsk: support for Tx") Fixes: bbff2f321a86 ("xsk: new descriptor addressing scheme") Fixes: 2b43470add8c ("xsk: Introduce AF_XDP buffer allocation API") Fixes: 26062b185eee ("xsk: Explicitly inline functions and move definitions") Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com> --- net/xdp/xsk_queue.h | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-)