wifi: iwlegacy: 3945-rs: fix possible null-pointer dereferences in il3945_rs_get_rate()
PATCH v2 1/1 wifi: iwlegacy: 3945-rs: fix possible null-pointer dereferences in il3945_rs_get_rate()
In this function, il_sta is assigned to rs_sta, and rs_sta is dereferenced at several points. If il_sta is NULL, this can lead to null-pointer dereferences. To fix this issue, add an early check for il_sta and return if it is NULL, consistent with the handling in il3945_rs_tx_status().
Besides, if the STA il data is uninitialized, return early instead of setting il_sta to NULL, consistent with the handling in il3945_rs_tx_status().
---
v2:
* Return early for uninitialized STA il data and align D_RATE messages with il3945_rs_tx_status(). Add a wifi: prefix to the patch title. Thanks to Stanislaw Gruszka for the helpful advice.
--- drivers/net/wireless/intel/iwlegacy/3945-rs.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-)
drivers/net/wireless/intel/iwlegacy/3945-rs.c +8 -3
@@ -626,10 +626,15 @@ il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta,D_RATE ("enter\n" ); +if (!il_sta) { +D_RATE ("leave: No STA il data to update!\n" ); +return ; + } +/* Treat uninitialized rate scaling data same as non-existing. */ -if (rs_sta && !rs_sta->il) { -D_RATE ("Rate scaling information not initialized yet.\n" ); - il_sta =NULL ; +if (!rs_sta->il) { +D_RATE ("leave: STA il data uninitialized!\n" ); +return ; } rate_mask = sta->deflink.supp_rates[sband->band];
Signature · 1 line
2.43.0
Quoted 10 lines · Tuo Li
In this function,
il_stais assigned tors_sta, andrs_stais dereferenced at several points. Ifil_stais NULL, this can lead to null-pointer dereferences. To fix this issue, add an early check foril_staand return if it is NULL, consistent with the handling inil3945_rs_tx_status().Besides, if the STA il data is uninitialized, return early instead of setting
il_stato NULL, consistent with the handling inil3945_rs_tx_status().Signed-off-by Tuo Li <[email protected]>
Thanks
Stanislaw
Quoted 35 lines
---
v2:
* Return early for uninitialized STA il data and alignD_RATEmessages withil3945_rs_tx_status(). Add a wifi: prefix to the patch title. Thanks to Stanislaw Gruszka for the helpful advice.--- drivers/net/wireless/intel/iwlegacy/3945-rs.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-)drivers/net/wireless/intel/iwlegacy/3945-rs.c +8
-3@@ -626,10 +626,15 @@ il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta,D_RATE ("enter\n" ); +if (!il_sta) { +D_RATE ("leave: No STA il data to update!\n" ); +return ; + } +/* Treat uninitialized rate scaling data same as non-existing. */ -if (rs_sta && !rs_sta->il) { -D_RATE ("Rate scaling information not initialized yet.\n" ); - il_sta =NULL ; +if (!rs_sta->il) { +D_RATE ("leave: STA il data uninitialized!\n" ); +return ; } rate_mask = sta->deflink.supp_rates[sband->band];Signature · 1 line
2.43.0
Quoted 6 lines · Stanislaw Gruszka
Quoted 5 lines · Tuo Li
In this function,
il_stais assigned tors_sta, andrs_stais dereferenced at several points. Ifil_stais NULL, this can lead to null-pointer dereferences. To fix this issue, add an early check foril_staand return if it is NULL, consistent with the handling inil3945_rs_tx_status().
Quoted 6 lines
Quoted 5 lines
Besides, if the STA il data is uninitialized, return early instead of setting
il_stato NULL, consistent with the handling inil3945_rs_tx_status().Signed-off-by Tuo Li <[email protected]>
I can apply this if you want, but for the record,
Quoted 10 lines
drivers/net/wireless/intel/iwlegacy/3945-rs.c il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta, · line 633
Quoted 10 lines
+++
b/drivers/net/wireless/intel/iwlegacy/3945-rs.cdrivers/net/wireless/intel/iwlegacy/3945-rs.c +5
-0@@ -626,10 +626,15 @@ il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta,D_RATE ("enter\n" ); +if (!il_sta) { +D_RATE ("leave: No STA il data to update!\n" ); +return ; + } +
I don't see how this would be possible. _Maybe_ the other one, but I can't figure out any scenario in mac80211 where it could happen either.
johannes
Hi Johannes,
Quoted 16 lines · Johannes Berg
I can apply this if you want, but for the record,
Quoted 10 lines
drivers/net/wireless/intel/iwlegacy/3945-rs.c il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta, · line 633
Quoted 10 lines
+++
b/drivers/net/wireless/intel/iwlegacy/3945-rs.cdrivers/net/wireless/intel/iwlegacy/3945-rs.c +5
-0@@ -626,10 +626,15 @@ il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta,D_RATE ("enter\n" ); +if (!il_sta) { +D_RATE ("leave: No STA il data to update!\n" ); +return ; + } +I don't see how this would be possible. _Maybe_ the other one, but I can't figure out any scenario in mac80211 where it could happen either.
Quoted 1 line
johannes
Thanks for the clarification.
I don't have a concrete mac80211 execution path that would result in il_sta being NULL here either. This issue was reported by a static analysis tool, and after reviewing the code I noticed that the handling is not consistent with il3945_rs_tx_status(), which is why I submitted this patch to add a defensive check.
If you believe this situation cannot occur in practice and the additional check is unnecessary, I'm fine with dropping this change.
Thanks for taking the time to review this.
Best regards,
Tuo
Stanislaw Gruszka
maintainerQuoted 8 lines · Tuo Li
Quoted 7 lines · Johannes Berg
I can apply this if you want, but for the record,
Quoted 5 lines
Quoted 5 lines
+++
b/drivers/net/wireless/intel/iwlegacy/3945-rs.cdrivers/net/wireless/intel/iwlegacy/3945-rs.c +0
-0@@ -626,10 +626,15 @@ il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta,D_RATE ("enter\n" );
Quoted 8 lines
Quoted 8 lines
Quoted 5 lines
drivers/net/wireless/intel/iwlegacy/3945-rs.c il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta, · line 633
Quoted 5 lines
drivers/net/wireless/intel/iwlegacy/3945-rs.c +5
-0+if (!il_sta) { +D_RATE ("leave: No STA il data to update!\n" ); +return ; + } +I don't see how this would be possible. _Maybe_ the other one, but I can't figure out any scenario in mac80211 where it could happen either.
Regarding checking the rs_sta->il, we can get rid of the ->il backpointer, it's only used for printing debug messages in a few functions. I don't think person needing to debug 3945 rate scaling algorithm exist nowadays :-)
I'll send patch for that.
Quoted 5 lines
I don't have a concrete mac80211 execution path that would result in
il_stabeing NULL here either. This issue was reported by a static analysis tool, and after reviewing the code I noticed that the handling is not consistent withil3945_rs_tx_status(), which is why I submitted this patch to add a defensive check.
IMO is ok to have defensive checks (in reasonable amount :-)
They can be marked with WARN_ON_ONCE like this:
if (WARN_ON_ONCE(!il_sta))
return
that would clearly indicate the check is for 'not possible' scenario.
Regards
Stanislaw
Quoted 36 lines · Stanislaw Gruszka
Quoted 8 lines · Tuo Li
Quoted 7 lines · Johannes Berg
I can apply this if you want, but for the record,
Quoted 5 lines
Quoted 5 lines
+++
b/drivers/net/wireless/intel/iwlegacy/3945-rs.cdrivers/net/wireless/intel/iwlegacy/3945-rs.c +0
-0@@ -626,10 +626,15 @@ il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta,D_RATE ("enter\n" );Quoted 8 lines
Quoted 8 lines
Quoted 5 lines
drivers/net/wireless/intel/iwlegacy/3945-rs.c il3945_rs_get_rate(void *il_r, struct ieee80211_sta *sta, void *il_sta, · line 633
Quoted 5 lines
drivers/net/wireless/intel/iwlegacy/3945-rs.c +5
-0+if (!il_sta) { +D_RATE ("leave: No STA il data to update!\n" ); +return ; + } +I don't see how this would be possible. _Maybe_ the other one, but I can't figure out any scenario in mac80211 where it could happen either.
Regarding checking the
rs_sta->il, we can get rid of the ->il backpointer, it's only used for printing debug messages in a few functions. I don't think person needing to debug 3945 rate scaling algorithm exist nowadays :-)I'll send patch for that.
Quoted 5 lines
I don't have a concrete mac80211 execution path that would result in
il_stabeing NULL here either. This issue was reported by a static analysis tool, and after reviewing the code I noticed that the handling is not consistent withil3945_rs_tx_status(), which is why I submitted this patch to add a defensive check.IMO is ok to have defensive checks (in reasonable amount :-)
They can be marked with
WARN_ON_ONCElike this:
Quoted 9 lines
if (WARN_ON_ONCE(!il_sta)) returnthat would clearly indicate the check is for 'not possible' scenario.
Regards
Stanislaw
Hi Stanislaw,
Thanks for your reply.
I will add a defensive WARN_ON_ONCE() at the beginning of il3945_rs_get_rate() to catch this unexpected condition, and will submit a v3 patch accordingly.
Best regards,
Tuo