Re: [PATCH] IB/core: Temporarily disable ex_query_device uverb

Hi,

Le lundi 02 février 2015 à 14:10 +0100, Yann Droneaud a écrit :

Le dimanche 01 février 2015 à 15:35 +0200, Haggai Eran a écrit :

quoted

Commit 5a77abf9a97a ("IB/core: Add support for extended query device caps")
added a new extended verb to query the capabilities of RDMA devices, but the
semantics of this verb are still under debate [1].

Block access to this verb from user-space until the new semantics are in
place.

Cc: Yann Droneaud <redacted>
Cc: Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
Cc: Eli Cohen <eli-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

[1] [PATCH v1 0/5] IB/core: extended query device caps cleanup for v3.19
    http://www.spinics.net/lists/linux-rdma/msg22904.html

Signed-off-by: Haggai Eran <haggaie-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

Reviewed-by: Yann Droneaud <redacted>

quoted

 drivers/infiniband/core/uverbs_main.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c
index e6c23b9eab33..5db1a8cc388d 100644
--- a/drivers/infiniband/core/uverbs_main.c
+++ b/drivers/infiniband/core/uverbs_main.c

@@ -123,7 +123,6 @@ static int (*uverbs_ex_cmd_table[])(struct ib_uverbs_file *file,
 				    struct ib_udata *uhw) = {
 	[IB_USER_VERBS_EX_CMD_CREATE_FLOW]	= ib_uverbs_ex_create_flow,
 	[IB_USER_VERBS_EX_CMD_DESTROY_FLOW]	= ib_uverbs_ex_destroy_flow,
-	[IB_USER_VERBS_EX_CMD_QUERY_DEVICE]	= ib_uverbs_ex_query_device
 };
 
 static void ib_uverbs_add_one(struct ib_device *device);

That's the smallest (and smartest) patch to be applied instead of
reverting.

Unfortunately, I've missed the issue I was complaining about in the
first place [1]. And I feel a bit guilty having missed the issue.

The present patch is fine as it fully disable the new extended 
QUERY_DEVICE uverb, but it doesn't disable the broken logic added in
ib_copy_to_udata() by commit 5a77abf9a97a ('IB/core: Add support for
extended query device caps'):

  diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
  index 470a011d6fa4..97a999f9e4d8 100644
  --- a/include/rdma/ib_verbs.h
  +++ b/include/rdma/ib_verbs.h
  @@ -1662,7 +1662,10 @@ static inline int ib_copy_from_udata(void *dest, struct ib_udata *udata, size_t
   
   static inline int ib_copy_to_udata(struct ib_udata *udata, void *src, size_t len)
   {
  -	return copy_to_user(udata->outbuf, src, len) ? -EFAULT : 0;
  +	size_t copy_sz;
  +
  +	copy_sz = min_t(size_t, len, udata->outlen);
  +	return copy_to_user(udata->outbuf, src, copy_sz) ? -EFAULT : 0;
   }

That part of commit 5a77abf9a97a should be reverted as I'm not sure
it doesn't introduce regressions, especially difficult to notice ones.

Regards.

[1] Re: [PATCH v3 06/17] IB/core: Add support for extended query device caps
    http://mid.gmane.org/1418733236.2779.26.camel-RlY5vtjFyJ3QT0dZR+AlfA@public.gmane.org

Regards.

-- 
Yann Droneaud
OPTEYA