net/9p: fix response size check in p9_check_errors()
Subsystems (2)
Christian Schoenebeck
reviewerv2 cover net/9p: fix response size check in p9_check_errors()
Follow-up fix for:
https://lore.kernel.org/linux-kernel/[email protected]/
Dominique, I moved zc to the end of the structure to somewhat address the layout padding. If you prefer a function argument in the first place, let me know and I'll send a v3.
v1 -> v2:
* Move zc to end of struct p9_rstatf (to avoid huge padding).
[patch 1]
* Fix format specifier for capacity (reported by kernel test bot).
[patch 2]
* Fix code style. [patch 1..2]
* Extend and adjust commit log messages. [patch 1..2]
Christian Schoenebeck (2):
net/9p: distinguish zero-copy requests net/9p: fix response size check in p9_check_errors()
include/net/9p/9p.h | 2 ++ net/9p/client.c | 13 +++++++++---- 2 files changed, 11 insertions(+), 4 deletions(-)
Signature · 1 line
2.30.2
Christian Schoenebeck
reviewerPATCH v2 1/2 net/9p: distinguish zero-copy requests
Add boolean zc member to struct p9_fcall to distinguish zero-copy messages (not using the linear sdata buffer for message payload) from regular messages (which do copy message payload to sdata before being further processed).
This new member is appended to end of structure to avoid inserting huge padding in generated layout.
--- include/net/9p/9p.h | 2 ++ net/9p/client.c | 6 ++++++ 2 files changed, 8 insertions(+)
include/net/9p/9p.h +2 -0
@@ -531,6 +531,7 @@ struct p9_rstatfs {* @offset: used by marshalling routines to track current position in buffer * @capacity: used by marshalling routines to track total malloc'd capacity * @sdata: payload +* @zc: whether zero-copy is used * * &p9_fcall represents the structure for all 9P RPC * transactions. Requests are packaged into fcalls, and reponses
@@ -549,6 +550,7 @@ struct p9_fcall {struct kmem_cache *cache;u8 *sdata; +bool zc; };int p9_errstr2errno (char *errstr,int len);
net/9p/client.c +6 -0
@@ -680,6 +680,9 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...)if (IS_ERR (req))return req; + req->tc.zc = false; + req->rc.zc = false; +if (signal_pending (current)) { sigpending =1 ;clear_thread_flag (TIF_SIGPENDING );
@@ -778,6 +781,9 @@ static struct p9_req_t *p9_client_zc_rpc(struct p9_client *c, int8_t type,if (IS_ERR (req))return req; + req->tc.zc = true; + req->rc.zc = true; +if (signal_pending (current)) { sigpending =1 ;clear_thread_flag (TIF_SIGPENDING );
Signature · 1 line
2.30.2
Christian Schoenebeck
reviewerPATCH v2 2/2 net/9p: fix response size check in p9_check_errors()
In reply to Christian SchoenebeckSince commit 60ece0833b6c ("net/9p: allocate appropriate reduced message buffers") it is no longer appropriate to check server's response size against msize. Check against the previously allocated buffer capacity instead.
- Omit this size check entirely for zero-copy messages, as those always allocate 4k (P9_ZC_HDR_SZ) linear buffers which are not used for actual payload and can be much bigger than 4k.
- Replace p9_debug() by pr_err() to make sure this message is always printed in case this error is triggered.
- Add 9p message type to error message to ease investigation.
--- net/9p/client.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-)
net/9p/client.c +3 -4
@@ -514,10 +514,9 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req)int ecode; err =p9_parse_header (&req->rc,NULL , &type,NULL ,0 ); -if (req->rc.size >= c->msize) { -p9_debug (P9_DEBUG_ERROR , -"requested packet size too big: %d\n" , - req->rc.size); +if (req->rc.size > req->rc.capacity && !req->rc.zc) { +pr_err ("requested packet size too big: %d does not fit %zu (type=%d)\n" , + req->rc.size, req->rc.capacity, req->rc.id);return -EIO ; }/* dump the response from server
Signature · 1 line
2.30.2