The TAP driver does not initialize all the elements of the rte_flow structure. This can lead to crash in rte_flow_destroy. (gdb) where flow=0x100e99280, error=0x0) at drivers/net/tap/tap_flow.c:1514 (gdb) p remote_flow $1 = (struct rte_flow *) 0x6b6b6b6b6b6b6b6b Which is here: static int tap_flow_destroy_pmd(struct pmd_internals *pmd, struct rte_flow *flow, struct rte_flow_error *error) { struct rte_flow *remote_flow = flow->remote_flow; ... if (remote_flow) { remote_flow->msg.nh.nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK; Simplest fix is to use rte_zmalloc() so remote_flow and other fields are always set at zero. Fixes: 2bc06869cd94 ("net/tap: add remote netdevice traffic capture") Cc: pascal.mazon@6wind.com Signed-off-by: Stephen Hemminger <stephen@networkplumber.org> --- drivers/net/tap/tap_flow.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/net/tap/tap_flow.c b/drivers/net/tap/tap_flow.c index 9d90361d9924..1538349e9c92 100644 --- a/drivers/net/tap/tap_flow.c +++ b/drivers/net/tap/tap_flow.c @@ -1380,7 +1380,7 @@ tap_flow_create(struct rte_eth_dev *dev, NULL, "priority value too big"); goto fail; } - flow = rte_malloc(__func__, sizeof(struct rte_flow), 0); + flow = rte_zmalloc(__func__, sizeof(struct rte_flow), 0); if (!flow) { rte_flow_error_set(error, ENOMEM, RTE_FLOW_ERROR_TYPE_HANDLE, NULL, "cannot allocate memory for rte_flow"); @@ -1416,7 +1416,7 @@ tap_flow_create(struct rte_eth_dev *dev, * to the local pmd->if_index. */ if (pmd->remote_if_index) { - remote_flow = rte_malloc(__func__, sizeof(struct rte_flow), 0); + remote_flow = rte_zmalloc(__func__, sizeof(struct rte_flow), 0); if (!remote_flow) { rte_flow_error_set( error, ENOMEM, RTE_FLOW_ERROR_TYPE_HANDLE, NULL, @@ -1693,7 +1693,7 @@ int tap_flow_implicit_create(struct pmd_internals *pmd, } }; - remote_flow = rte_malloc(__func__, sizeof(struct rte_flow), 0); + remote_flow = rte_zmalloc(__func__, sizeof(struct rte_flow), 0); if (!remote_flow) { TAP_LOG(ERR, "Cannot allocate memory for rte_flow"); goto fail; @@ -1896,7 +1896,7 @@ static int rss_enable(struct pmd_internals *pmd, return -ENOTSUP; } - rss_flow = rte_malloc(__func__, sizeof(struct rte_flow), 0); + rss_flow = rte_zmalloc(__func__, sizeof(struct rte_flow), 0); if (!rss_flow) { TAP_LOG(ERR, "Cannot allocate memory for rte_flow"); -- 2.20.1
On Mon, Apr 27, 2020 at 11:39 PM Stephen Hemminger
<stephen@networkplumber.org> wrote:
>
> The TAP driver does not initialize all the elements of the rte_flow
> structure. This can lead to crash in rte_flow_destroy.
>
> (gdb) where
> flow=0x100e99280, error=0x0)
> at drivers/net/tap/tap_flow.c:1514
>
> (gdb) p remote_flow
> $1 = (struct rte_flow *) 0x6b6b6b6b6b6b6b6b
>
> Which is here:
> static int
> tap_flow_destroy_pmd(struct pmd_internals *pmd,
> struct rte_flow *flow,
> struct rte_flow_error *error)
> {
> struct rte_flow *remote_flow = flow->remote_flow;
> ...
> if (remote_flow) {
> remote_flow->msg.nh.nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK;
>
> Simplest fix is to use rte_zmalloc() so remote_flow and other fields
> are always set at zero.
>
> Fixes: 2bc06869cd94 ("net/tap: add remote netdevice traffic capture")
> Cc: pascal.mazon@6wind.com
Not sure why you copied Pascal (I'd say he stopped working on dpdk 2 years ago).
Please use the devtools/get-maintainer.sh script.
Thanks.
--
David Marchand
On 4/27/2020 10:39 PM, Stephen Hemminger wrote: > The TAP driver does not initialize all the elements of the rte_flow > structure. This can lead to crash in rte_flow_destroy. > > (gdb) where > flow=0x100e99280, error=0x0) > at drivers/net/tap/tap_flow.c:1514 > > (gdb) p remote_flow > $1 = (struct rte_flow *) 0x6b6b6b6b6b6b6b6b > > Which is here: > static int > tap_flow_destroy_pmd(struct pmd_internals *pmd, > struct rte_flow *flow, > struct rte_flow_error *error) > { > struct rte_flow *remote_flow = flow->remote_flow; > ... > if (remote_flow) { > remote_flow->msg.nh.nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK; > > Simplest fix is to use rte_zmalloc() so remote_flow and other fields > are always set at zero. Both 'rte_malloc' & 'rte_zmalloc' should be zeroing the allocated memory, unless MALLOC_DEBUG config option set [1], if this is not the case the issue can be still valid after this change. [1] http://lxr.dpdk.org/dpdk/v20.02/source/lib/librte_eal/common/rte_malloc.c#L83 > > Fixes: 2bc06869cd94 ("net/tap: add remote netdevice traffic capture") > Cc: pascal.mazon@6wind.com > Signed-off-by: Stephen Hemminger <stephen@networkplumber.org> <...>
On Fri, 1 May 2020 17:01:40 +0100
Ferruh Yigit <ferruh.yigit@intel.com> wrote:
> On 4/27/2020 10:39 PM, Stephen Hemminger wrote:
> > The TAP driver does not initialize all the elements of the rte_flow
> > structure. This can lead to crash in rte_flow_destroy.
> >
> > (gdb) where
> > flow=0x100e99280, error=0x0)
> > at drivers/net/tap/tap_flow.c:1514
> >
> > (gdb) p remote_flow
> > $1 = (struct rte_flow *) 0x6b6b6b6b6b6b6b6b
> >
> > Which is here:
> > static int
> > tap_flow_destroy_pmd(struct pmd_internals *pmd,
> > struct rte_flow *flow,
> > struct rte_flow_error *error)
> > {
> > struct rte_flow *remote_flow = flow->remote_flow;
> > ...
> > if (remote_flow) {
> > remote_flow->msg.nh.nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK;
> >
> > Simplest fix is to use rte_zmalloc() so remote_flow and other fields
> > are always set at zero.
>
> Both 'rte_malloc' & 'rte_zmalloc' should be zeroing the allocated memory, unless
> MALLOC_DEBUG config option set [1], if this is not the case the issue can be
> still valid after this change.
Malloc debug poisons memory to find bugs like this.
On 5/1/2020 5:19 PM, Stephen Hemminger wrote:
> On Fri, 1 May 2020 17:01:40 +0100
> Ferruh Yigit <ferruh.yigit@intel.com> wrote:
>
>> On 4/27/2020 10:39 PM, Stephen Hemminger wrote:
>>> The TAP driver does not initialize all the elements of the rte_flow
>>> structure. This can lead to crash in rte_flow_destroy.
>>>
>>> (gdb) where
>>> flow=0x100e99280, error=0x0)
>>> at drivers/net/tap/tap_flow.c:1514
>>>
>>> (gdb) p remote_flow
>>> $1 = (struct rte_flow *) 0x6b6b6b6b6b6b6b6b
>>>
>>> Which is here:
>>> static int
>>> tap_flow_destroy_pmd(struct pmd_internals *pmd,
>>> struct rte_flow *flow,
>>> struct rte_flow_error *error)
>>> {
>>> struct rte_flow *remote_flow = flow->remote_flow;
>>> ...
>>> if (remote_flow) {
>>> remote_flow->msg.nh.nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK;
>>>
>>> Simplest fix is to use rte_zmalloc() so remote_flow and other fields
>>> are always set at zero.
>>
>> Both 'rte_malloc' & 'rte_zmalloc' should be zeroing the allocated memory, unless
>> MALLOC_DEBUG config option set [1], if this is not the case the issue can be
>> still valid after this change.
>
> Malloc debug poisons memory to find bugs like this.
>
Fair enough, if that is the intention use 'rte_zmalloc' to clarify it.
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
On 5/5/2020 9:09 AM, Ferruh Yigit wrote:
> On 5/1/2020 5:19 PM, Stephen Hemminger wrote:
>> On Fri, 1 May 2020 17:01:40 +0100
>> Ferruh Yigit <ferruh.yigit@intel.com> wrote:
>>
>>> On 4/27/2020 10:39 PM, Stephen Hemminger wrote:
>>>> The TAP driver does not initialize all the elements of the rte_flow
>>>> structure. This can lead to crash in rte_flow_destroy.
>>>>
>>>> (gdb) where
>>>> flow=0x100e99280, error=0x0)
>>>> at drivers/net/tap/tap_flow.c:1514
>>>>
>>>> (gdb) p remote_flow
>>>> $1 = (struct rte_flow *) 0x6b6b6b6b6b6b6b6b
>>>>
>>>> Which is here:
>>>> static int
>>>> tap_flow_destroy_pmd(struct pmd_internals *pmd,
>>>> struct rte_flow *flow,
>>>> struct rte_flow_error *error)
>>>> {
>>>> struct rte_flow *remote_flow = flow->remote_flow;
>>>> ...
>>>> if (remote_flow) {
>>>> remote_flow->msg.nh.nlmsg_flags = NLM_F_REQUEST | NLM_F_ACK;
>>>>
>>>> Simplest fix is to use rte_zmalloc() so remote_flow and other fields
>>>> are always set at zero.
>>>
>>> Both 'rte_malloc' & 'rte_zmalloc' should be zeroing the allocated memory, unless
>>> MALLOC_DEBUG config option set [1], if this is not the case the issue can be
>>> still valid after this change.
>>
>> Malloc debug poisons memory to find bugs like this.
>>
>
> Fair enough, if that is the intention use 'rte_zmalloc' to clarify it.
>
> Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
>
Applied to dpdk-next-net/master, thanks.