Skip to contentSkip to conversation

net: mvneta: Prevent out of bounds read in mvneta_config_rss()

PATCH Applied Dan Carpenter 3 participants 6 messages Dec 2, 2022 – Dec 5, 2022

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net: mvneta: Prevent out of bounds read in mvneta_config_rss()

The pp->indir[0] value comes from the user. It is passed to:

	if (cpu_online(pp->rxq_def))

inside the mvneta_percpu_elect() function. It needs bounds checkeding to ensure that it is not beyond the end of the cpu bitmap.

Fixes cad5d847a093 ("net: mvneta: Fix the CPU choice in mvneta_percpu_elect")
---
 drivers/net/ethernet/marvell/mvneta.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/net/ethernet/marvell/mvneta.c +3 -0
index c2cb98d24f5c..5abc7c3e399e 100644
@@ -4927,6 +4927,9 @@ static int  mvneta_config_rss(struct mvneta_port *pp)
 		napi_disable(&pp->napi);
 	}
 
+	if (pp->indir[0] >= nr_cpu_ids)
+		return -EINVAL;
+
 	pp->rxq_def = pp->indir[0];
 
 	/* Update unicast mapping */
Signature · 1 line
2.35.1
#1 lore raw
Quoted 4 lines · Dan Carpenter

The pp->indir[0] value comes from the user. It is passed to:

	if (cpu_online(pp->rxq_def))
Quoted 8 lines

inside the mvneta_percpu_elect() function. It needs bounds checkeding to ensure that it is not beyond the end of the cpu bitmap.

Fixes cad5d847a093 ("net: mvneta: Fix the CPU choice in mvneta_percpu_elect")
---
 drivers/net/ethernet/marvell/mvneta.c | 3 +++
 1 file changed, 3 insertions(+)

I would expect that ethtool_copy_validate_indir() will prevent this.

Thanks

Quoted 18 lines
drivers/net/ethernet/marvell/mvneta.c +3 -0
index c2cb98d24f5c..5abc7c3e399e 100644
@@ -4927,6 +4927,9 @@ static int  mvneta_config_rss(struct mvneta_port *pp)
 		napi_disable(&pp->napi);
 	}
 
+	if (pp->indir[0] >= nr_cpu_ids)
+		return -EINVAL;
+
 	pp->rxq_def = pp->indir[0];
 
 	/* Update unicast mapping */
Signature · 1 line
2.35.1
#2 lore raw
Quoted 8 lines · Leon Romanovsky
Quoted 7 lines · Dan Carpenter

The pp->indir[0] value comes from the user. It is passed to:

	if (cpu_online(pp->rxq_def))

inside the mvneta_percpu_elect() function. It needs bounds checkeding to ensure that it is not beyond the end of the cpu bitmap.

Quoted 8 lines
Quoted 5 lines
Fixes cad5d847a093 ("net: mvneta: Fix the CPU choice in mvneta_percpu_elect")
---
 drivers/net/ethernet/marvell/mvneta.c | 3 +++
 1 file changed, 3 insertions(+)

I would expect that ethtool_copy_validate_indir() will prevent this.

Huh... Sort of, but in the strictest sense, no. mvneta_ethtool_get_rxnfc() sets the cap at 8 by default or an unvalidated module parameter.

regards,
dan carpenter

#5 lore raw
In reply to Dan Carpenter
This patch was applied to netdev/net.git (master) by David S. Miller <[email protected]>:

Hello:

This patch was applied to netdev/net.git (master)
by David S. Miller <[email protected]>:

Quoted 8 lines · On Fri, 2 Dec 2022 12:58:26 +0300 you wrote:

The pp->indir[0] value comes from the user. It is passed to:

	if (cpu_online(pp->rxq_def))

inside the mvneta_percpu_elect() function. It needs bounds checkeding to ensure that it is not beyond the end of the cpu bitmap.

[...]

Here is the summary with links:
- [net] net: mvneta: Prevent out of bounds read in mvneta_config_rss()

    https://git.kernel.org/netdev/net/c/e8b4fc13900b

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
#3 lore raw
In reply to Dan Carpenter
Quoted 15 lines · Dan Carpenter
Quoted 14 lines · Leon Romanovsky
Quoted 12 lines · Dan Carpenter

The pp->indir[0] value comes from the user. It is passed to:

	if (cpu_online(pp->rxq_def))

inside the mvneta_percpu_elect() function. It needs bounds checkeding to ensure that it is not beyond the end of the cpu bitmap.

Fixes cad5d847a093 ("net: mvneta: Fix the CPU choice in mvneta_percpu_elect")
---
 drivers/net/ethernet/marvell/mvneta.c | 3 +++
 1 file changed, 3 insertions(+)
Quoted 5 lines
Quoted 2 lines

I would expect that ethtool_copy_validate_indir() will prevent this.

Huh... Sort of, but in the strictest sense, no. mvneta_ethtool_get_rxnfc() sets the cap at 8 by default or an unvalidated module parameter.

And is this solely mvnet issue? Do other drivers safe for this input?

Thanks

Quoted 4 lines

regards,
dan carpenter

#4 lore raw
Quoted 16 lines · Leon Romanovsky
Quoted 15 lines · Dan Carpenter
Quoted 14 lines · Leon Romanovsky
Quoted 12 lines · Dan Carpenter

The pp->indir[0] value comes from the user. It is passed to:

	if (cpu_online(pp->rxq_def))

inside the mvneta_percpu_elect() function. It needs bounds checkeding to ensure that it is not beyond the end of the cpu bitmap.

Fixes cad5d847a093 ("net: mvneta: Fix the CPU choice in mvneta_percpu_elect")
---
 drivers/net/ethernet/marvell/mvneta.c | 3 +++
 1 file changed, 3 insertions(+)
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

I would expect that ethtool_copy_validate_indir() will prevent this.

Huh... Sort of, but in the strictest sense, no. mvneta_ethtool_get_rxnfc() sets the cap at 8 by default or an unvalidated module parameter.

And is this solely mvnet issue? Do other drivers safe for this input?

I believe so, yes. However thinking about it now maybe a better fix would be to go back to the original way of using pp->rxq_def % nr_cpu_ids. (Originally it used num_online_cpus() instead of nr_cpu_ids but I think nr_cpu_ids is correct). I will send this patch tomorrow.

In this code, if you hit the out of bounds then you kind of deserve it, but there are probably a lot of people who probably have fewer than 8 cores and in that case the bug results in a WARN().

regards,
dan carpenter