* [PATCH] [v2] qed: Fix a potential use-after-free in qed_cxt_tables_alloc
@ 2023-12-07 9:36 Dinghao Liu
2023-12-07 9:55 ` Przemek Kitszel
2023-12-08 23:59 ` Jakub Kicinski
0 siblings, 2 replies; 4+ messages in thread
From: Dinghao Liu @ 2023-12-07 9:36 UTC (permalink / raw)
To: dinghao.liu
Cc: Ariel Elior, Manish Chopra, David S. Miller, Eric Dumazet,
Jakub Kicinski, Paolo Abeni, Yuval Mintz, netdev, linux-kernel
qed_ilt_shadow_alloc() will call qed_ilt_shadow_free() to
free p_hwfn->p_cxt_mngr->ilt_shadow on error. However,
qed_cxt_tables_alloc() accesses the freed pointer on failure
of qed_ilt_shadow_alloc() through calling qed_cxt_mngr_free(),
which may lead to use-after-free. Fix this issue by setting
p_hwfn->p_cxt_mngr->ilt_shadow to NULL in qed_ilt_shadow_free().
Fixes: fe56b9e6a8d9 ("qed: Add module with basic common support")
Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn>
---
Changelog:
v2: -Change the bug type from double-free to use-after-free.
-Move the null check against p_mngr->ilt_shadow to the beginning
of the function qed_ilt_shadow_free().
-When kcalloc() fails in qed_ilt_shadow_alloc(), just return
because there is nothing to free.
---
drivers/net/ethernet/qlogic/qed/qed_cxt.c | 12 +++++++-----
1 file changed, 7 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/qlogic/qed/qed_cxt.c b/drivers/net/ethernet/qlogic/qed/qed_cxt.c
index 65e20693c549..911e0c0d3563 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_cxt.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_cxt.c
@@ -921,9 +921,12 @@ static void qed_ilt_shadow_free(struct qed_hwfn *p_hwfn)
struct qed_cxt_mngr *p_mngr = p_hwfn->p_cxt_mngr;
u32 ilt_size, i;
+ if (!p_mngr->ilt_shadow)
+ return;
+
ilt_size = qed_cxt_ilt_shadow_size(p_cli);
- for (i = 0; p_mngr->ilt_shadow && i < ilt_size; i++) {
+ for (i = 0; i < ilt_size; i++) {
struct phys_mem_desc *p_dma = &p_mngr->ilt_shadow[i];
if (p_dma->virt_addr)
@@ -933,6 +936,7 @@ static void qed_ilt_shadow_free(struct qed_hwfn *p_hwfn)
p_dma->virt_addr = NULL;
}
kfree(p_mngr->ilt_shadow);
+ p_hwfn->p_cxt_mngr->ilt_shadow = NULL;
}
static int qed_ilt_blk_alloc(struct qed_hwfn *p_hwfn,
@@ -995,10 +999,8 @@ static int qed_ilt_shadow_alloc(struct qed_hwfn *p_hwfn)
size = qed_cxt_ilt_shadow_size(clients);
p_mngr->ilt_shadow = kcalloc(size, sizeof(struct phys_mem_desc),
GFP_KERNEL);
- if (!p_mngr->ilt_shadow) {
- rc = -ENOMEM;
- goto ilt_shadow_fail;
- }
+ if (!p_mngr->ilt_shadow)
+ return -ENOMEM;
DP_VERBOSE(p_hwfn, QED_MSG_ILT,
"Allocated 0x%x bytes for ilt shadow\n",
--
2.17.1
^ permalink raw reply related [flat|nested] 4+ messages in thread* Re: [PATCH] [v2] qed: Fix a potential use-after-free in qed_cxt_tables_alloc
2023-12-07 9:36 [PATCH] [v2] qed: Fix a potential use-after-free in qed_cxt_tables_alloc Dinghao Liu
@ 2023-12-07 9:55 ` Przemek Kitszel
2023-12-08 23:59 ` Jakub Kicinski
1 sibling, 0 replies; 4+ messages in thread
From: Przemek Kitszel @ 2023-12-07 9:55 UTC (permalink / raw)
To: Dinghao Liu
Cc: Ariel Elior, Manish Chopra, David S. Miller, Eric Dumazet,
Jakub Kicinski, Paolo Abeni, Yuval Mintz, netdev, linux-kernel
On 12/7/23 10:36, Dinghao Liu wrote:
> qed_ilt_shadow_alloc() will call qed_ilt_shadow_free() to
> free p_hwfn->p_cxt_mngr->ilt_shadow on error. However,
> qed_cxt_tables_alloc() accesses the freed pointer on failure
> of qed_ilt_shadow_alloc() through calling qed_cxt_mngr_free(),
> which may lead to use-after-free. Fix this issue by setting
> p_hwfn->p_cxt_mngr->ilt_shadow to NULL in qed_ilt_shadow_free().
>
> Fixes: fe56b9e6a8d9 ("qed: Add module with basic common support")
> Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn>
> ---
>
> Changelog:
For future submissions please also provide links to previous versions
(would be to v1 in this case). No need to add this now for this one.
>
> v2: -Change the bug type from double-free to use-after-free.
> -Move the null check against p_mngr->ilt_shadow to the beginning
> of the function qed_ilt_shadow_free().
> -When kcalloc() fails in qed_ilt_shadow_alloc(), just return
> because there is nothing to free.
> ---
> drivers/net/ethernet/qlogic/qed/qed_cxt.c | 12 +++++++-----
> 1 file changed, 7 insertions(+), 5 deletions(-)
>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] [v2] qed: Fix a potential use-after-free in qed_cxt_tables_alloc
2023-12-07 9:36 [PATCH] [v2] qed: Fix a potential use-after-free in qed_cxt_tables_alloc Dinghao Liu
2023-12-07 9:55 ` Przemek Kitszel
@ 2023-12-08 23:59 ` Jakub Kicinski
2023-12-09 12:40 ` dinghao.liu
1 sibling, 1 reply; 4+ messages in thread
From: Jakub Kicinski @ 2023-12-08 23:59 UTC (permalink / raw)
To: Dinghao Liu
Cc: Ariel Elior, Manish Chopra, David S. Miller, Eric Dumazet,
Paolo Abeni, Yuval Mintz, netdev, linux-kernel
On Thu, 7 Dec 2023 17:36:06 +0800 Dinghao Liu wrote:
> v2: -Change the bug type from double-free to use-after-free.
> -Move the null check against p_mngr->ilt_shadow to the beginning
> of the function qed_ilt_shadow_free().
> -When kcalloc() fails in qed_ilt_shadow_alloc(), just return
> because there is nothing to free.
This refactoring is not acceptable as part of a fix, sorry.
> @@ -933,6 +936,7 @@ static void qed_ilt_shadow_free(struct qed_hwfn *p_hwfn)
> p_dma->virt_addr = NULL;
> }
> kfree(p_mngr->ilt_shadow);
> + p_hwfn->p_cxt_mngr->ilt_shadow = NULL;
Why do you dereference p_hwfn here?
Seems more natural to use:
p_mngr->ilt_shadow = NULL;
since that's the exact pointer that was passed to free.
--
pw-bot: cr
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] [v2] qed: Fix a potential use-after-free in qed_cxt_tables_alloc
2023-12-08 23:59 ` Jakub Kicinski
@ 2023-12-09 12:40 ` dinghao.liu
0 siblings, 0 replies; 4+ messages in thread
From: dinghao.liu @ 2023-12-09 12:40 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Ariel Elior, Manish Chopra, David S. Miller, Eric Dumazet,
Paolo Abeni, Yuval Mintz, netdev, linux-kernel
> On Thu, 7 Dec 2023 17:36:06 +0800 Dinghao Liu wrote:
> > v2: -Change the bug type from double-free to use-after-free.
> > -Move the null check against p_mngr->ilt_shadow to the beginning
> > of the function qed_ilt_shadow_free().
> > -When kcalloc() fails in qed_ilt_shadow_alloc(), just return
> > because there is nothing to free.
>
> This refactoring is not acceptable as part of a fix, sorry.
>
> > @@ -933,6 +936,7 @@ static void qed_ilt_shadow_free(struct qed_hwfn *p_hwfn)
> > p_dma->virt_addr = NULL;
> > }
> > kfree(p_mngr->ilt_shadow);
> > + p_hwfn->p_cxt_mngr->ilt_shadow = NULL;
>
> Why do you dereference p_hwfn here?
> Seems more natural to use:
>
> p_mngr->ilt_shadow = NULL;
>
> since that's the exact pointer that was passed to free.
> --
> pw-bot: cr
I will resend a new patch to fix this, thanks!
Regards,
Dinghao
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2023-12-09 12:41 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-12-07 9:36 [PATCH] [v2] qed: Fix a potential use-after-free in qed_cxt_tables_alloc Dinghao Liu
2023-12-07 9:55 ` Przemek Kitszel
2023-12-08 23:59 ` Jakub Kicinski
2023-12-09 12:40 ` dinghao.liu
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).