rbd: fix cleanup in rbd_add()
Subsystems (2)
PATCH 1/1 rbd: fix cleanup in rbd_add()
Bjorn Helgaas pointed out that a recent commit introduced a use-after-free condition in an error path for rbd_add(). He correctly stated:
I think b536f69a3a5 "rbd: set up devices only for mapped images" introduced a use-after-free error in rbd_add():
...
If rbd_dev_device_setup() returns an error, we call rbd_dev_image_release(), which ultimately kfrees rbd_dev. Then we call rbd_dev_destroy(), which references fields in the already-freed rbd_dev struct before kfreeing it again.
The simple fix is to return the error code after the call to rbd_dev_image_release().
Closer examination revealed that there's no need to clean up ceph_opts or rbd_opts in that function, so fix that too.
Update some other comments that have also become out of date.
--- drivers/block/rbd.c | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 deletions(-)
drivers/block/rbd.c +4 -2
@@ -4699,8 +4699,10 @@ out:return ret; } -/* Undo whatever state changes are made by v1 or v2 image probe */ - +/* +* Undo whatever state changes are made by v1 or v2 header info +* call. +*/ static void rbd_dev_unprobe (struct rbd_device *rbd_dev) {struct rbd_image_header *header;
@@ -4904,9 +4906,10 @@ static int rbd_dev_image_probe(struct rbd_device
*rbd_dev, bool mapping)
int tmp;
/*
- * Get the id from the image id object. If it's not a
- * format 2 image, we'll get ENOENT back, and we'll assume
- * it's a format 1 image.
+ * Get the id from the image id object. Unless there's an
+ * error, rbd_dev->spec->image_id will be filled in with
+ * a dynamically-allocated string, and rbd_dev->image_format
+ * will be set to either 1 or 2.
*/
ret = rbd_dev_image_id(rbd_dev);
if (ret)
@@ -4994,7 +4997,7 @@ static ssize_t rbd_add(struct bus_type *bus,
rc = PTR_ERR(rbdc);
goto err_out_args;
}
- ceph_opts = NULL; /* rbd_dev client now owns this */
+ ceph_opts = NULL; /* rbd_get_client() consumes this */
/* pick the pool */
osdc = &rbdc->client->osdc;
@@ -5033,14 +5036,14 @@ static ssize_t rbd_add(struct bus_type *bus,
return count;
rbd_dev_image_release(rbd_dev);
+
+ return rc;
+
err_out_rbd_dev:
rbd_dev_destroy(rbd_dev);
err_out_client:
rbd_put_client(rbdc);
err_out_args:
- if (ceph_opts)
- ceph_destroy_options(ceph_opts);
- kfree(rbd_opts);
rbd_spec_put(spec);
err_out_module:
module_put(THIS_MODULE);
Signature · 1 line
1.7.9.5
PATCH v2 1/1 rbd: fix cleanup in rbd_add()
Bjorn Helgaas pointed out that a recent commit introduced a use-after-free condition in an error path for rbd_add(). He correctly stated:
I think b536f69a3a5 "rbd: set up devices only for mapped images" introduced a use-after-free error in rbd_add():
...
If rbd_dev_device_setup() returns an error, we call rbd_dev_image_release(), which ultimately kfrees rbd_dev. Then we call rbd_dev_destroy(), which references fields in the already-freed rbd_dev struct before kfreeing it again.
The simple fix is to return the error code after the call to rbd_dev_image_release().
Closer examination revealed that there's no need to clean up rbd_opts in that function, so fix that too.
Update some other comments that have also become out of date.
---
v2: call to ceph_destroy_options() moved to its own patch
drivers/block/rbd.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-)
drivers/block/rbd.c +4 -2
@@ -4700,8 +4700,10 @@ out:return ret; } -/* Undo whatever state changes are made by v1 or v2 image probe */ - +/* +* Undo whatever state changes are made by v1 or v2 header info +* call. +*/ static void rbd_dev_unprobe (struct rbd_device *rbd_dev) {struct rbd_image_header *header;
@@ -4905,9 +4907,10 @@ static int rbd_dev_image_probe(struct rbd_device
*rbd_dev, bool mapping)
int tmp;
/*
- * Get the id from the image id object. If it's not a
- * format 2 image, we'll get ENOENT back, and we'll assume
- * it's a format 1 image.
+ * Get the id from the image id object. Unless there's an
+ * error, rbd_dev->spec->image_id will be filled in with
+ * a dynamically-allocated string, and rbd_dev->image_format
+ * will be set to either 1 or 2.
*/
ret = rbd_dev_image_id(rbd_dev);
if (ret)
@@ -5033,12 +5036,14 @@ static ssize_t rbd_add(struct bus_type *bus,
return count;
rbd_dev_image_release(rbd_dev);
+
+ return rc;
+
err_out_rbd_dev:
rbd_dev_destroy(rbd_dev);
err_out_client:
rbd_put_client(rbdc);
err_out_args:
- kfree(rbd_opts);
rbd_spec_put(spec);
err_out_module:
module_put(THIS_MODULE);
Signature · 1 line
1.7.9.5
Quoted 61 lines · Alex Elder
Bjorn Helgaas pointed out that a recent commit introduced a use-after-free condition in an error path for
rbd_add(). He correctly stated:I think
b536f69a3a5"rbd: set up devices only for mapped images" introduced a use-after-free error inrbd_add():
...
Ifrbd_dev_device_setup()returns an error, we callrbd_dev_image_release(), which ultimately kfreesrbd_dev. Then we callrbd_dev_destroy(), which references fields in the already-freedrbd_devstruct before kfreeing it again.The simple fix is to return the error code after the call to
rbd_dev_image_release().Closer examination revealed that there's no need to clean up
rbd_optsin that function, so fix that too.Update some other comments that have also become out of date.
Reported-by Bjorn Helgaas <[email protected]>Signed-off-by Alex Elder <[email protected]>---
v2: call toceph_destroy_options()moved to its own patchdrivers/block/rbd.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-)drivers/block/rbd.c +8
-5@@ -4700,8 +4700,10 @@ out:return ret; } -/* Undo whatever state changes are made by v1 or v2 image probe */ - +/* +* Undo whatever state changes are made by v1 or v2 header info +* call. +*/ static void rbd_dev_unprobe (struct rbd_device *rbd_dev) {struct rbd_image_header *header;@@ -4905,9 +4907,10 @@ static int rbd_dev_image_probe(struct rbd_device *rbd_dev, bool mapping)int tmp;/* -* Get the id from the image id object. If it's not a -* format 2 image, we'll get ENOENT back, and we'll assume -* it's a format 1 image. +* Get the id from the image id object. Unless there's an +* error, rbd_dev->spec->image_id will be filled in with +* a dynamically-allocated string, and rbd_dev->image_format +* will be set to either 1 or 2. */ ret =rbd_dev_image_id (rbd_dev);if (ret)
drivers/block/rbd.c static ssize_t rbd_add(struct bus_type *bus, · line 5038
Quoted 3 lines
drivers/block/rbd.c +0
-0@@ -5033,12 +5036,14 @@ static ssize_t rbd_add(struct bus_type *bus,return count;rbd_dev_image_release (rbd_dev);
It looks like rbd_dev_image_release() does all the needed cleanup except the module_put(). Just adding a module_put() here would do the trick I think, since rbd_dev_image_release() is also used for parent images that don't call module_get() and module_put().
With that change:
Quoted 12 lines
drivers/block/rbd.c +3
-1+ +return rc; + err_out_rbd_dev:rbd_dev_destroy (rbd_dev); err_out_client:rbd_put_client (rbdc); err_out_args: -kfree (rbd_opts);rbd_spec_put (spec); err_out_module:module_put (THIS_MODULE );
Quoted 64 lines · Josh Durgin
Quoted 63 lines · Alex Elder
Bjorn Helgaas pointed out that a recent commit introduced a use-after-free condition in an error path for
rbd_add(). He correctly stated:I think
b536f69a3a5"rbd: set up devices only for mapped images" introduced a use-after-free error inrbd_add():
...
Ifrbd_dev_device_setup()returns an error, we callrbd_dev_image_release(), which ultimately kfreesrbd_dev. Then we callrbd_dev_destroy(), which references fields in the already-freedrbd_devstruct before kfreeing it again.The simple fix is to return the error code after the call to
rbd_dev_image_release().Closer examination revealed that there's no need to clean up
rbd_optsin that function, so fix that too.Update some other comments that have also become out of date.
Reported-by Bjorn Helgaas <[email protected]>Signed-off-by Alex Elder <[email protected]>---
v2: call toceph_destroy_options()moved to its own patchdrivers/block/rbd.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-)drivers/block/rbd.c +8
-5@@ -4700,8 +4700,10 @@ out:return ret; } -/* Undo whatever state changes are made by v1 or v2 image probe */ - +/* +* Undo whatever state changes are made by v1 or v2 header info +* call. +*/ static void rbd_dev_unprobe (struct rbd_device *rbd_dev) {struct rbd_image_header *header;@@ -4905,9 +4907,10 @@ static int rbd_dev_image_probe(struct rbd_device *rbd_dev, bool mapping)int tmp;/* -* Get the id from the image id object. If it's not a -* format 2 image, we'll get ENOENT back, and we'll assume -* it's a format 1 image. +* Get the id from the image id object. Unless there's an +* error, rbd_dev->spec->image_id will be filled in with +* a dynamically-allocated string, and rbd_dev->image_format +* will be set to either 1 or 2. */ ret =rbd_dev_image_id (rbd_dev);if (ret)
drivers/block/rbd.c +0
-0@@ -5033,12 +5036,14 @@ static ssize_t rbd_add(struct bus_type *bus,return count;
Quoted 6 lines
drivers/block/rbd.c static ssize_t rbd_add(struct bus_type *bus, · line 5038
Quoted 1 line
drivers/block/rbd.c +0
-0rbd_dev_image_release (rbd_dev);It looks like
rbd_dev_image_release()does all the needed cleanup except themodule_put(). Just adding amodule_put()here would do the trick I think, sincerbd_dev_image_release()is also used for parent images that don't callmodule_get()andmodule_put().
It took me a bit to understand that what you're really saying is that this path is missing a module_put() call...
(But that seems to be the only thing missing.)
I'll fix that. Actually, I may just make it
goto err_out_module instead, so we get the same
debug message when there's a problem.
-Alex
Quoted 17 lines
With that change:
Reviewed-by Josh Durgin <[email protected]>Quoted 12 lines
drivers/block/rbd.c +3
-1+ +return rc; + err_out_rbd_dev:rbd_dev_destroy (rbd_dev); err_out_client:rbd_put_client (rbdc); err_out_args: -kfree (rbd_opts);rbd_spec_put (spec); err_out_module:module_put (THIS_MODULE );