2022-07-23 04:12:35

by Duoming Zhou

Subject: [PATCH net] netrom: fix sleep in atomic context bugs in timer handlers

There are sleep in atomic context bugs in timer handlers of netrom
such as nr_t1timer_expiry(), nr_t2timer_expiry(), nr_heartbeat_expiry(),
nr_idletimer_expiry() and so on.

The root cause is kmemdup() with GFP_KERNEL parameter that may sleep
could be called by different timer handlers which is in interrupt context.

One of the call paths that could trigger bug is shown below:

(interrupt context)
nr_heartbeat_expiry
nr_write_internal
nr_transmit_buffer
nr_route_frame
nr_add_node
kmemdup(..,GFP_KERNEL) //may sleep

This patch changes gfp_t parameter of kmemdup in nr_add_node()
from GFP_KERNEL to GFP_ATOMIC in order to prevent sleep in atomic
context bugs.

Fixes: eafff86d3bd8 ("[NETROM]: Use kmemdup")
Signed-off-by: Duoming Zhou <[email protected]>
---
net/netrom/nr_route.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/netrom/nr_route.c b/net/netrom/nr_route.c
index baea3cbd76c..1ddcf13de6a 100644
--- a/net/netrom/nr_route.c
+++ b/net/netrom/nr_route.c
@@ -163,7 +163,7 @@ static int __must_check nr_add_node(ax25_address *nr, const char *mnemonic,
if (ax25_digi != NULL && ax25_digi->ndigi > 0) {
nr_neigh->digipeat = kmemdup(ax25_digi,
sizeof(*ax25_digi),
- GFP_KERNEL);
+ GFP_ATOMIC);
if (nr_neigh->digipeat == NULL) {
kfree(nr_neigh);
if (nr_node)
--
2.17.1

2022-07-26 02:52:25

by Jakub Kicinski

Subject: Re: [PATCH net] netrom: fix sleep in atomic context bugs in timer handlers

On Sat, 23 Jul 2022 11:56:46 +0800 Duoming Zhou wrote:
> Fixes: eafff86d3bd8 ("[NETROM]: Use kmemdup")

That's not a correct Fixes tag, acme just swapped kmalloc for kmemdup().
The allocate flags did not change.

2022-07-26 03:26:19

by Duoming Zhou

Subject: Re: [PATCH net] netrom: fix sleep in atomic context bugs in timer handlers

Hello,

On Mon, 25 Jul 2022 19:49:30 -0700 Jakub Kicinski wrote:

> On Sat, 23 Jul 2022 11:56:46 +0800 Duoming Zhou wrote:
> > Fixes: eafff86d3bd8 ("[NETROM]: Use kmemdup")
>
> That's not a correct Fixes tag, acme just swapped kmalloc for kmemdup().
> The allocate flags did not change.

Thanks for your time and reply!

The correct Fixes tag is "Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")".
The following is the code:

https://elixir.bootlin.com/linux/v2.6.12-rc2/source/net/netrom/nr_route.c#L158

...
if ((nr_neigh->digipeat = kmalloc(sizeof(*ax25_digi), GFP_KERNEL)) == NULL) {
...

Best regards,
Duoming Zhou