* [PATCH] fsl/fman: Validate cell-index value obtained from Device Tree
@ 2024-07-02 9:50 Aleksandr Mishin
2024-07-02 13:36 ` Simon Horman
2024-07-02 14:01 ` [PATCH v2] " Aleksandr Mishin
0 siblings, 2 replies; 5+ messages in thread
From: Aleksandr Mishin @ 2024-07-02 9:50 UTC (permalink / raw)
To: Igal Liberman
Cc: Aleksandr Mishin, Madalin Bucur, Sean Anderson, David S. Miller,
Eric Dumazet, Jakub Kicinski, Paolo Abeni, netdev, linux-kernel,
lvc-project
Cell-index value is obtained from Device Tree and then used to calculate
the index for accessing arrays port_mfl[], mac_mfl[] and intr_mng[].
In case of broken DT due to any error cell-index can contain any value
and it is possible to go beyond the array boundaries which can lead
at least to memory corruption.
Validate cell-index value obtained from Device Tree.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Fixes: 414fd46e7762 ("fsl/fman: Add FMan support")
Signed-off-by: Aleksandr Mishin <amishin@t-argos.ru>
---
drivers/net/ethernet/freescale/fman/fman.c | 7 +++++++
drivers/net/ethernet/freescale/fman/fman.h | 2 ++
drivers/net/ethernet/freescale/fman/mac.c | 5 +++++
3 files changed, 14 insertions(+)
diff --git a/drivers/net/ethernet/freescale/fman/fman.c b/drivers/net/ethernet/freescale/fman/fman.c
index d96028f01770..6929bca3f768 100644
--- a/drivers/net/ethernet/freescale/fman/fman.c
+++ b/drivers/net/ethernet/freescale/fman/fman.c
@@ -2933,3 +2933,10 @@ module_exit(fman_unload);
MODULE_LICENSE("Dual BSD/GPL");
MODULE_DESCRIPTION("Freescale DPAA Frame Manager driver");
+
+int check_mac_id(u32 mac_id)
+{
+ if (mac_id >= MAX_NUM_OF_MACS)
+ return -EINVAL;
+ return 0;
+}
diff --git a/drivers/net/ethernet/freescale/fman/fman.h b/drivers/net/ethernet/freescale/fman/fman.h
index 2ea575a46675..3cedde4851e1 100644
--- a/drivers/net/ethernet/freescale/fman/fman.h
+++ b/drivers/net/ethernet/freescale/fman/fman.h
@@ -372,6 +372,8 @@ u16 fman_get_max_frm(void);
int fman_get_rx_extra_headroom(void);
+int check_mac_id(u32 mac_id);
+
#ifdef CONFIG_DPAA_ERRATUM_A050385
bool fman_has_errata_a050385(void);
#endif
diff --git a/drivers/net/ethernet/freescale/fman/mac.c b/drivers/net/ethernet/freescale/fman/mac.c
index 9767586b4eb3..7a67b4c887e2 100644
--- a/drivers/net/ethernet/freescale/fman/mac.c
+++ b/drivers/net/ethernet/freescale/fman/mac.c
@@ -247,6 +247,11 @@ static int mac_probe(struct platform_device *_of_dev)
dev_err(dev, "failed to read cell-index for %pOF\n", mac_node);
return -EINVAL;
}
+ err = check_mac_id(val);
+ if (err) {
+ dev_err(dev, "cell-index value is out of range for %pOF\n", mac_node);
+ return err;
+ }
priv->cell_index = (u8)val;
/* Get the MAC address */
--
2.30.2
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH] fsl/fman: Validate cell-index value obtained from Device Tree
2024-07-02 9:50 [PATCH] fsl/fman: Validate cell-index value obtained from Device Tree Aleksandr Mishin
@ 2024-07-02 13:36 ` Simon Horman
2024-07-02 17:44 ` Jakub Kicinski
2024-07-02 14:01 ` [PATCH v2] " Aleksandr Mishin
1 sibling, 1 reply; 5+ messages in thread
From: Simon Horman @ 2024-07-02 13:36 UTC (permalink / raw)
To: Aleksandr Mishin
Cc: Igal Liberman, Madalin Bucur, Sean Anderson, David S. Miller,
Eric Dumazet, Jakub Kicinski, Paolo Abeni, netdev, linux-kernel,
lvc-project
On Tue, Jul 02, 2024 at 12:50:34PM +0300, Aleksandr Mishin wrote:
> Cell-index value is obtained from Device Tree and then used to calculate
> the index for accessing arrays port_mfl[], mac_mfl[] and intr_mng[].
> In case of broken DT due to any error cell-index can contain any value
> and it is possible to go beyond the array boundaries which can lead
> at least to memory corruption.
> Validate cell-index value obtained from Device Tree.
>
> Found by Linux Verification Center (linuxtesting.org) with SVACE.
>
> Fixes: 414fd46e7762 ("fsl/fman: Add FMan support")
> Signed-off-by: Aleksandr Mishin <amishin@t-argos.ru>
> ---
> drivers/net/ethernet/freescale/fman/fman.c | 7 +++++++
> drivers/net/ethernet/freescale/fman/fman.h | 2 ++
> drivers/net/ethernet/freescale/fman/mac.c | 5 +++++
> 3 files changed, 14 insertions(+)
>
> diff --git a/drivers/net/ethernet/freescale/fman/fman.c b/drivers/net/ethernet/freescale/fman/fman.c
> index d96028f01770..6929bca3f768 100644
> --- a/drivers/net/ethernet/freescale/fman/fman.c
> +++ b/drivers/net/ethernet/freescale/fman/fman.c
> @@ -2933,3 +2933,10 @@ module_exit(fman_unload);
>
> MODULE_LICENSE("Dual BSD/GPL");
> MODULE_DESCRIPTION("Freescale DPAA Frame Manager driver");
> +
> +int check_mac_id(u32 mac_id)
> +{
> + if (mac_id >= MAX_NUM_OF_MACS)
> + return -EINVAL;
> + return 0;
> +}
> diff --git a/drivers/net/ethernet/freescale/fman/fman.h b/drivers/net/ethernet/freescale/fman/fman.h
> index 2ea575a46675..3cedde4851e1 100644
> --- a/drivers/net/ethernet/freescale/fman/fman.h
> +++ b/drivers/net/ethernet/freescale/fman/fman.h
> @@ -372,6 +372,8 @@ u16 fman_get_max_frm(void);
>
> int fman_get_rx_extra_headroom(void);
>
> +int check_mac_id(u32 mac_id);
> +
> #ifdef CONFIG_DPAA_ERRATUM_A050385
> bool fman_has_errata_a050385(void);
> #endif
> diff --git a/drivers/net/ethernet/freescale/fman/mac.c b/drivers/net/ethernet/freescale/fman/mac.c
> index 9767586b4eb3..7a67b4c887e2 100644
> --- a/drivers/net/ethernet/freescale/fman/mac.c
> +++ b/drivers/net/ethernet/freescale/fman/mac.c
> @@ -247,6 +247,11 @@ static int mac_probe(struct platform_device *_of_dev)
> dev_err(dev, "failed to read cell-index for %pOF\n", mac_node);
> return -EINVAL;
> }
> + err = check_mac_id(val);
Hi Aleksandr, all,
It seems that this breaks linking with allmodconfig builds on x86_64,
perhaps it is better to simply make check_mac_id a static function in mac.c ?
> + if (err) {
> + dev_err(dev, "cell-index value is out of range for %pOF\n", mac_node);
Although other instances exist in this function before this patch,
it does look like this leaks a reference to of_dev taken by the
call to of_find_device_by_node() on line 194.
Maybe it is intentional, I'm unsure.
Perhaps this can be investigated separately to the fix proposed by this
patch?
Flagged by Coccinelle (this one is on line 253):
.../mac.c:238:2-8: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:242:2-8: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:248:2-8: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:253:2-8: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:267:2-8: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:273:2-8: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:282:3-9: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:320:2-8: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:333:1-7: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:337:1-7: ERROR: missing put_device; call of_find_device_by_node on line 194, but without a corresponding object release within this function.
.../mac.c:282:3-9: ERROR: missing put_device; call of_find_device_by_node on line 285, but without a corresponding object release within this function.
.../mac.c:320:2-8: ERROR: missing put_device; call of_find_device_by_node on line 285, but without a corresponding object release within this function.
.../mac.c:333:1-7: ERROR: missing put_device; call of_find_device_by_node on line 285, but without a corresponding object release within this function.
.../mac.c:337:1-7: ERROR: missing put_device; call of_find_device_by_node on line 285, but without a corresponding object release within this function.
> + return err;
> + }
> priv->cell_index = (u8)val;
>
> /* Get the MAC address */
--
pw-bot: changes-requested
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH v2] fsl/fman: Validate cell-index value obtained from Device Tree
2024-07-02 9:50 [PATCH] fsl/fman: Validate cell-index value obtained from Device Tree Aleksandr Mishin
2024-07-02 13:36 ` Simon Horman
@ 2024-07-02 14:01 ` Aleksandr Mishin
2024-07-02 14:50 ` Sean Anderson
1 sibling, 1 reply; 5+ messages in thread
From: Aleksandr Mishin @ 2024-07-02 14:01 UTC (permalink / raw)
To: Igal Liberman
Cc: Aleksandr Mishin, Madalin Bucur, Sean Anderson, David S. Miller,
Eric Dumazet, Jakub Kicinski, Paolo Abeni, netdev, linux-kernel,
lvc-project
Cell-index value is obtained from Device Tree and then used to calculate
the index for accessing arrays port_mfl[], mac_mfl[] and intr_mng[].
In case of broken DT due to any error cell-index can contain any value
and it is possible to go beyond the array boundaries which can lead
at least to memory corruption.
Validate cell-index value obtained from Device Tree.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Fixes: 414fd46e7762 ("fsl/fman: Add FMan support")
Signed-off-by: Aleksandr Mishin <amishin@t-argos.ru>
---
v1->v2: Move check to mac.c to avoid allmodconfig build errors and reference leaks
drivers/net/ethernet/freescale/fman/fman.c | 1 -
drivers/net/ethernet/freescale/fman/fman.h | 3 +++
drivers/net/ethernet/freescale/fman/mac.c | 4 ++++
3 files changed, 7 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/freescale/fman/fman.c b/drivers/net/ethernet/freescale/fman/fman.c
index d96028f01770..fb416d60dcd7 100644
--- a/drivers/net/ethernet/freescale/fman/fman.c
+++ b/drivers/net/ethernet/freescale/fman/fman.c
@@ -24,7 +24,6 @@
/* General defines */
#define FMAN_LIODN_TBL 64 /* size of LIODN table */
-#define MAX_NUM_OF_MACS 10
#define FM_NUM_OF_FMAN_CTRL_EVENT_REGS 4
#define BASE_RX_PORTID 0x08
#define BASE_TX_PORTID 0x28
diff --git a/drivers/net/ethernet/freescale/fman/fman.h b/drivers/net/ethernet/freescale/fman/fman.h
index 2ea575a46675..74eb62eba0d7 100644
--- a/drivers/net/ethernet/freescale/fman/fman.h
+++ b/drivers/net/ethernet/freescale/fman/fman.h
@@ -74,6 +74,9 @@
#define BM_MAX_NUM_OF_POOLS 64 /* Buffers pools */
#define FMAN_PORT_MAX_EXT_POOLS_NUM 8 /* External BM pools per Rx port */
+/* General defines */
+#define MAX_NUM_OF_MACS 10
+
struct fman; /* FMan data */
/* Enum for defining port types */
diff --git a/drivers/net/ethernet/freescale/fman/mac.c b/drivers/net/ethernet/freescale/fman/mac.c
index 9767586b4eb3..ac9ad5e67b44 100644
--- a/drivers/net/ethernet/freescale/fman/mac.c
+++ b/drivers/net/ethernet/freescale/fman/mac.c
@@ -247,6 +247,10 @@ static int mac_probe(struct platform_device *_of_dev)
dev_err(dev, "failed to read cell-index for %pOF\n", mac_node);
return -EINVAL;
}
+ if (val >= MAX_NUM_OF_MACS) {
+ dev_err(dev, "cell-index value is too big for %pOF\n", mac_node);
+ return -EINVAL;
+ }
priv->cell_index = (u8)val;
/* Get the MAC address */
--
2.30.2
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH v2] fsl/fman: Validate cell-index value obtained from Device Tree
2024-07-02 14:01 ` [PATCH v2] " Aleksandr Mishin
@ 2024-07-02 14:50 ` Sean Anderson
0 siblings, 0 replies; 5+ messages in thread
From: Sean Anderson @ 2024-07-02 14:50 UTC (permalink / raw)
To: Aleksandr Mishin, Igal Liberman
Cc: Madalin Bucur, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, netdev, linux-kernel, lvc-project
On 7/2/24 10:01, Aleksandr Mishin wrote:
> [You don't often get email from amishin@t-argos.ru. Learn why this is important at https://cas5-0-urlprotect.trendmicro.com:443/wis/clicktime/v1/query?url=https%3a%2f%2faka.ms%2fLearnAboutSenderIdentification&umid=ca782a5b-a3fd-4738-b50c-ffbd8e52e50f&auth=d807158c60b7d2502abde8a2fc01f40662980862-ac0deac714bbe85d389fd2711a17d0113034b3ca ]
>
> Cell-index value is obtained from Device Tree and then used to calculate
> the index for accessing arrays port_mfl[], mac_mfl[] and intr_mng[].
> In case of broken DT due to any error cell-index can contain any value
> and it is possible to go beyond the array boundaries which can lead
> at least to memory corruption.
> Validate cell-index value obtained from Device Tree.
>
> Found by Linux Verification Center (linuxtesting.org) with SVACE.
>
> Fixes: 414fd46e7762 ("fsl/fman: Add FMan support")
> Signed-off-by: Aleksandr Mishin <amishin@t-argos.ru>
> ---
> v1->v2: Move check to mac.c to avoid allmodconfig build errors and reference leaks
>
> drivers/net/ethernet/freescale/fman/fman.c | 1 -
> drivers/net/ethernet/freescale/fman/fman.h | 3 +++
> drivers/net/ethernet/freescale/fman/mac.c | 4 ++++
> 3 files changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/ethernet/freescale/fman/fman.c b/drivers/net/ethernet/freescale/fman/fman.c
> index d96028f01770..fb416d60dcd7 100644
> --- a/drivers/net/ethernet/freescale/fman/fman.c
> +++ b/drivers/net/ethernet/freescale/fman/fman.c
> @@ -24,7 +24,6 @@
>
> /* General defines */
> #define FMAN_LIODN_TBL 64 /* size of LIODN table */
> -#define MAX_NUM_OF_MACS 10
> #define FM_NUM_OF_FMAN_CTRL_EVENT_REGS 4
> #define BASE_RX_PORTID 0x08
> #define BASE_TX_PORTID 0x28
> diff --git a/drivers/net/ethernet/freescale/fman/fman.h b/drivers/net/ethernet/freescale/fman/fman.h
> index 2ea575a46675..74eb62eba0d7 100644
> --- a/drivers/net/ethernet/freescale/fman/fman.h
> +++ b/drivers/net/ethernet/freescale/fman/fman.h
> @@ -74,6 +74,9 @@
> #define BM_MAX_NUM_OF_POOLS 64 /* Buffers pools */
> #define FMAN_PORT_MAX_EXT_POOLS_NUM 8 /* External BM pools per Rx port */
>
> +/* General defines */
> +#define MAX_NUM_OF_MACS 10
> +
> struct fman; /* FMan data */
>
> /* Enum for defining port types */
> diff --git a/drivers/net/ethernet/freescale/fman/mac.c b/drivers/net/ethernet/freescale/fman/mac.c
> index 9767586b4eb3..ac9ad5e67b44 100644
> --- a/drivers/net/ethernet/freescale/fman/mac.c
> +++ b/drivers/net/ethernet/freescale/fman/mac.c
> @@ -247,6 +247,10 @@ static int mac_probe(struct platform_device *_of_dev)
> dev_err(dev, "failed to read cell-index for %pOF\n", mac_node);
> return -EINVAL;
> }
> + if (val >= MAX_NUM_OF_MACS) {
> + dev_err(dev, "cell-index value is too big for %pOF\n", mac_node);
> + return -EINVAL;
> + }
> priv->cell_index = (u8)val;
>
> /* Get the MAC address */
> --
> 2.30.2
>
Reviewed-by: Sean Anderson <sean.anderson@seco.com>
[StudioX, SECO SpA]<https://www.seco.com/applications/studiox>
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] fsl/fman: Validate cell-index value obtained from Device Tree
2024-07-02 13:36 ` Simon Horman
@ 2024-07-02 17:44 ` Jakub Kicinski
0 siblings, 0 replies; 5+ messages in thread
From: Jakub Kicinski @ 2024-07-02 17:44 UTC (permalink / raw)
To: Aleksandr Mishin
Cc: Simon Horman, Igal Liberman, Madalin Bucur, Sean Anderson,
David S. Miller, Eric Dumazet, Paolo Abeni, netdev, linux-kernel,
lvc-project
On Tue, 2 Jul 2024 14:36:51 +0100 Simon Horman wrote:
> Maybe it is intentional, I'm unsure.
> Perhaps this can be investigated separately to the fix proposed by this
> patch?
+1, please fix this first
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2024-07-02 17:44 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-07-02 9:50 [PATCH] fsl/fman: Validate cell-index value obtained from Device Tree Aleksandr Mishin
2024-07-02 13:36 ` Simon Horman
2024-07-02 17:44 ` Jakub Kicinski
2024-07-02 14:01 ` [PATCH v2] " Aleksandr Mishin
2024-07-02 14:50 ` Sean Anderson
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).