From: Greg KH <gregkh@suse.de>
To: linux-kernel@vger.kernel.org, stable@kernel.org
Cc: stable-review@kernel.org, torvalds@linux-foundation.org,
akpm@linux-foundation.org, alan@lxorguk.ukuu.org.uk,
Sean Hefty <sean.hefty@intel.com>,
Doug Ledford <dledford@redhat.com>,
Roland Dreier <roland@purestorage.com>
Subject: [13/44] RDMA/cma: Fix crash in request handlers
Date: Mon, 21 Mar 2011 16:09:53 -0700 [thread overview]
Message-ID: <20110321231052.103742258@clark.kroah.org> (raw)
In-Reply-To: <20110321231105.GA1744@kroah.com>
2.6.32-longterm review patch. If anyone has any objections, please let us know.
------------------
From: Sean Hefty <sean.hefty@intel.com>
commit 25ae21a10112875763c18b385624df713a288a05 upstream.
Doug Ledford and Red Hat reported a crash when running the rdma_cm on
a real-time OS. The crash has the following call trace:
cm_process_work
cma_req_handler
cma_disable_callback
rdma_create_id
kzalloc
init_completion
cma_get_net_info
cma_save_net_info
cma_any_addr
cma_zero_addr
rdma_translate_ip
rdma_copy_addr
cma_acquire_dev
rdma_addr_get_sgid
ib_find_cached_gid
cma_attach_to_dev
ucma_event_handler
kzalloc
ib_copy_ah_attr_to_user
cma_comp
[ preempted ]
cma_write
copy_from_user
ucma_destroy_id
copy_from_user
_ucma_find_context
ucma_put_ctx
ucma_free_ctx
rdma_destroy_id
cma_exch
cma_cancel_operation
rdma_node_get_transport
rt_mutex_slowunlock
bad_area_nosemaphore
oops_enter
They were able to reproduce the crash multiple times with the
following details:
Crash seems to always happen on the:
mutex_unlock(&conn_id->handler_mutex);
as conn_id looks to have been freed during this code path.
An examination of the code shows that a race exists in the request
handlers. When a new connection request is received, the rdma_cm
allocates a new connection identifier. This identifier has a single
reference count on it. If a user calls rdma_destroy_id() from another
thread after receiving a callback, rdma_destroy_id will proceed to
destroy the id and free the associated memory. However, the request
handlers may still be in the process of running. When control returns
to the request handlers, they can attempt to access the newly created
identifiers.
Fix this by holding a reference on the newly created rdma_cm_id until
the request handler is through accessing it.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Acked-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
---
drivers/infiniband/core/cma.c | 15 +++++++++++++++
1 file changed, 15 insertions(+)
--- a/drivers/infiniband/core/cma.c
+++ b/drivers/infiniband/core/cma.c
@@ -1138,6 +1138,11 @@ static int cma_req_handler(struct ib_cm_
cm_id->context = conn_id;
cm_id->cm_handler = cma_ib_handler;
+ /*
+ * Protect against the user destroying conn_id from another thread
+ * until we're done accessing it.
+ */
+ atomic_inc(&conn_id->refcount);
ret = conn_id->id.event_handler(&conn_id->id, &event);
if (!ret) {
/*
@@ -1150,8 +1155,10 @@ static int cma_req_handler(struct ib_cm_
ib_send_cm_mra(cm_id, CMA_CM_MRA_SETTING, NULL, 0);
mutex_unlock(&lock);
mutex_unlock(&conn_id->handler_mutex);
+ cma_deref_id(conn_id);
goto out;
}
+ cma_deref_id(conn_id);
/* Destroy the CM ID by returning a non-zero value. */
conn_id->cm_id.ib = NULL;
@@ -1353,17 +1360,25 @@ static int iw_conn_req_handler(struct iw
event.param.conn.private_data_len = iw_event->private_data_len;
event.param.conn.initiator_depth = attr.max_qp_init_rd_atom;
event.param.conn.responder_resources = attr.max_qp_rd_atom;
+
+ /*
+ * Protect against the user destroying conn_id from another thread
+ * until we're done accessing it.
+ */
+ atomic_inc(&conn_id->refcount);
ret = conn_id->id.event_handler(&conn_id->id, &event);
if (ret) {
/* User wants to destroy the CM ID */
conn_id->cm_id.iw = NULL;
cma_exch(conn_id, CMA_DESTROYING);
mutex_unlock(&conn_id->handler_mutex);
+ cma_deref_id(conn_id);
rdma_destroy_id(&conn_id->id);
goto out;
}
mutex_unlock(&conn_id->handler_mutex);
+ cma_deref_id(conn_id);
out:
if (dev)
next prev parent reply other threads:[~2011-03-22 0:00 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-03-21 23:11 [00/44] 2.6.32.34-longterm review Greg KH
2011-03-21 23:09 ` [01/44] hwmon/f71882fg: Set platform drvdata to NULL later Greg KH
2011-03-21 23:09 ` [02/44] mtd: jedec_probe: Change variable name from cfi_p to cfi Greg KH
2011-03-22 1:58 ` [Stable-review] " Ben Hutchings
2011-03-22 22:25 ` Greg KH
2011-03-21 23:09 ` [03/44] mtd: add "platform:" prefix for platform modalias Greg KH
2011-03-21 23:09 ` [04/44] libata: no special completion processing for EH commands Greg KH
2011-03-21 23:09 ` [05/44] MIPS: MTX-1: Make au1000_eth probe all PHY addresses Greg KH
2011-03-21 23:09 ` [06/44] x86/mm: Handle mm_fault_error() in kernel space Greg KH
2011-03-21 23:09 ` [07/44] ftrace: Fix memory leak with function graph and cpu hotplug Greg KH
2011-03-21 23:09 ` [08/44] x86: Fix panic when handling "mem={invalid}" param Greg KH
2011-03-21 23:09 ` [09/44] x86: Emit "mem=nopentium ignored" warning when not supported Greg KH
2011-03-21 23:09 ` [10/44] ahci: AHCI and RAID mode SATA patch for Intel Patsburg DeviceIDs Greg KH
2011-03-21 23:09 ` [11/44] ahci: AHCI mode SATA patch for Intel DH89xxCC DeviceIDs Greg KH
2011-03-21 23:09 ` [12/44] ahci: AHCI mode SATA patch for Intel Patsburg SATA RAID controller Greg KH
2011-03-21 23:09 ` Greg KH [this message]
2011-03-21 23:09 ` [14/44] IB/cm: Bump reference count on cm_id before invoking callback Greg KH
2011-03-21 23:09 ` [15/44] x86, quirk: Fix SB600 revision check Greg KH
2011-03-21 23:09 ` [16/44] ath9k_hw: Fix incorrect macversion and macrev checks Greg KH
2011-03-21 23:09 ` [17/44] USB: serial/kobil_sct, fix potential tty NULL dereference Greg KH
2011-03-21 23:09 ` [18/44] USB: serial: ch341: add new id Greg KH
2011-03-21 23:09 ` [19/44] xhci: Fix cycle bit calculation during stall handling Greg KH
2011-03-21 23:10 ` [20/44] ALSA: hda - fix digital mic selection in mixer on 92HD8X codecs Greg KH
2011-03-21 23:10 ` [21/44] PCI: remove quirk for pre-production systems Greg KH
2011-03-21 23:10 ` [22/44] PCI: add more checking to ICH region quirks Greg KH
2011-03-21 23:10 ` [23/44] PCI: do not create quirk I/O regions below PCIBIOS_MIN_IO for ICH Greg KH
2011-03-21 23:10 ` [24/44] PCI: sysfs: Fix failure path for addition of "vpd" attribute Greg KH
2011-03-21 23:10 ` [25/44] ALSA: ctxfi - Fix incorrect SPDIF status bit mask Greg KH
2011-03-21 23:10 ` [26/44] ALSA: ctxfi - Fix SPDIF status retrieval Greg KH
2011-03-21 23:10 ` [27/44] ALSA: ctxfi - Clear input settings before initialization Greg KH
2011-03-21 23:10 ` [28/44] SUNRPC: Ensure we always run the tk_callback before tk_action Greg KH
2011-03-21 23:10 ` [29/44] perf: Handle stopped state with tracepoints Greg KH
2011-03-21 23:10 ` [30/44] perf, powerpc: Handle events that raise an exception without overflowing Greg KH
2011-03-21 23:10 ` [31/44] ext3: Always set dx_nodes fake_dirent explicitly Greg KH
2011-03-21 23:10 ` [32/44] call_function_many: fix list delete vs add race Greg KH
2011-03-21 23:10 ` [33/44] call_function_many: add missing ordering Greg KH
2011-03-21 23:10 ` [34/44] x86: Flush TLB if PGD entry is changed in i386 PAE mode Greg KH
2011-03-21 23:10 ` [35/44] isdn: avoid calling tty_ldisc_flush() in atomic context Greg KH
2011-03-21 23:10 ` [36/44] smp_call_function_many: handle concurrent clearing of mask Greg KH
2011-03-21 23:10 ` [37/44] [PARISC] fix per-cpu flag problem in the cpu affinity checkers Greg KH
2011-03-21 23:10 ` [38/44] i2c: Fix typo in instantiating-devices document Greg KH
2011-03-21 23:10 ` [39/44] mmc: sdio: remember new card RCA when redetecting card Greg KH
2011-03-21 23:10 ` [40/44] powerpc/kexec: Fix race in kexec shutdown Greg KH
2011-03-21 23:10 ` [41/44] powerpc/kdump: Fix race in kdump shutdown Greg KH
2011-03-21 23:10 ` [42/44] powerpc: rtas_flash needs to use rtas_data_buf Greg KH
2011-03-21 23:10 ` [43/44] x86, binutils, xen: Fix another wrong size directive Greg KH
2011-03-21 23:10 ` [44/44] hwmon: (sht15) Fix integer overflow in humidity calculation Greg KH
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20110321231052.103742258@clark.kroah.org \
--to=gregkh@suse.de \
--cc=akpm@linux-foundation.org \
--cc=alan@lxorguk.ukuu.org.uk \
--cc=dledford@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=roland@purestorage.com \
--cc=sean.hefty@intel.com \
--cc=stable-review@kernel.org \
--cc=stable@kernel.org \
--cc=torvalds@linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®