atm/fore200e: Fix possible data race in fore200e_open()
Subsystems (4)
PATCH v2 1/1 atm/fore200e: Fix possible data race in fore200e_open()
Protect access to fore200e->available_cell_rate with rate_mtx lock to prevent potential data race.
In this case, since the update depends on a prior read, a data race could lead to a wrong fore200e.available_cell_rate value.
The field fore200e.available_cell_rate is generally protected by the lock fore200e.rate_mtx when accessed. In all other read and write cases, this field is consistently protected by the lock, except for this case and during initialization.
This potential bug was detected by our experimental static analysis tool, which analyzes locking APIs and paired functions to identify data races and atomicity violations.
---
v2:
* Added a description of the data race hazard in fore200e_open(), as suggested by Jakub Kicinski and Simon Horman.
--- drivers/atm/fore200e.c | 2 ++ 1 file changed, 2 insertions(+)
drivers/atm/fore200e.c +2 -0
@@ -1374,7 +1374,9 @@ fore200e_open(struct atm_vcc *vcc) vcc->dev_data =NULL ; +mutex_lock (&fore200e->rate_mtx); fore200e->available_cell_rate += vcc->qos.txtp.max_pcr; +mutex_unlock (&fore200e->rate_mtx);kfree (fore200e_vcc);return -EINVAL ;
Signature · 1 line
2.25.1
Quoted 15 lines · Gui-Dong Han
Protect access to
fore200e->available_cell_ratewithrate_mtxlock to prevent potential data race.In this case, since the update depends on a prior read, a data race could lead to a wrong fore200e.available_cell_rate value.
The field fore200e.available_cell_rate is generally protected by the lock fore200e.rate_mtx when accessed. In all other read and write cases, this field is consistently protected by the lock, except for this case and during initialization.
This potential bug was detected by our experimental static analysis tool, which analyzes locking APIs and paired functions to identify data races and atomicity violations.
Quoted 7 lines
Fixes 1da177e4c3f4 ("Linux-2.6.12-rc2")Signed-off-by Gui-Dong Han <[email protected]>---
v2:
* Added a description of the data race hazard infore200e_open(), as suggested by Jakub Kicinski and Simon Horman.
Jakub Kicinski
maintainerQuoted 10 lines · On Wed, 22 Jan 2025 02:37:45 +0000 Gui-Dong Han wrote:
Protect access to
fore200e->available_cell_ratewithrate_mtxlock to prevent potential data race.In this case, since the update depends on a prior read, a data race could lead to a wrong fore200e.available_cell_rate value.
The field fore200e.available_cell_rate is generally protected by the lock fore200e.rate_mtx when accessed. In all other read and write cases, this field is consistently protected by the lock, except for this case and during initialization.
Please describe the call paths which interact to cause the race.
Quoted 8 lines · Jakub Kicinski
Quoted 6 lines · On Wed, 22 Jan 2025 02:37:45 +0000 Gui-Dong Han wrote:
Protect access to
fore200e->available_cell_ratewithrate_mtxlock to prevent potential data race.In this case, since the update depends on a prior read, a data race could lead to a wrong fore200e.available_cell_rate value.
Quoted 6 lines
Quoted 4 lines
The field fore200e.available_cell_rate is generally protected by the lock fore200e.rate_mtx when accessed. In all other read and write cases, this field is consistently protected by the lock, except for this case and during initialization.
Please describe the call paths which interact to cause the race.
The fore200e.available_cell_rate is a shared resource used to track the available bandwidth for allocation.
The functions fore200e_open(), fore200e_close(), and fore200e_change_qos(), which modify fore200e.available_cell_rate to reflect bandwidth availability, may interact and cause a race condition.
fore200e_open(struct atm_vcc *vcc)
{
...
/* Pseudo-CBR bandwidth requested? */
if ((vcc->qos.txtp.traffic_class == ATM_CBR) &&(vcc->qos.txtp.max_pcr > 0)) {
mutex_lock(&fore200e->rate_mtx);
if (fore200e->available_cell_rate < vcc->qos.txtp.max_pcr) {
mutex_unlock(&fore200e->rate_mtx);
... // Error handling code
return -EAGAIN;
}
/* Reserve bandwidth */
fore200e->available_cell_rate -= vcc->qos.txtp.max_pcr;
mutex_unlock(&fore200e->rate_mtx);
}
...
if (fore200e_activate_vcin(fore200e, 1, vcc, vcc->qos.rxtp.max_sdu) < 0) {
... // Error handling code
fore200e->available_cell_rate += vcc->qos.txtp.max_pcr;
... // Error handling code
return -EINVAL;
}
}There is further evidence within fore200e_open() itself. The function correctly takes the lock when subtracting vcc->qos.txtp.max_pcr from fore200e.available_cell_rate to reserve bandwidth, but later, in the error handling path, it adds vcc->qos.txtp.max_pcr back without holding the lock. There is no justification for modifying a shared resource without the corresponding lock in this case.
Regards,
Han
Hi Jakub and Simon,
I was organizing my emails and noticed this v2 patch from January. Simon kindly provided a "Reviewed-by" tag for it.
It seems this patch may have been overlooked and was not merged. I checked the current upstream tree, and the data race in fore200e_open() (accessing fore200e->available_cell_rate without the rate_mtx lock in the error path) still exists.
Could you please take another look and consider this patch for merging?
Thank you,
Gui-Dong Han
Simon Horman
reviewerQuoted 11 lines · Gui-Dong Han
Hi Jakub and Simon,
I was organizing my emails and noticed this v2 patch from January. Simon kindly provided a "Reviewed-by" tag for it.
It seems this patch may have been overlooked and was not merged. I checked the current upstream tree, and the data race in
fore200e_open()(accessingfore200e->available_cell_ratewithout therate_mtxlock in the error path) still exists.Could you please take another look and consider this patch for merging?
Hi,
January was a long time ago, so I guess this slipped through the cracks somehow.
I would suggest reposting it to have it considered for merging. I'd also explicitly target the patch at net. Something like this:
Subject: [PATCH REPOST net v2] ...
And feel free to include my Reviewed-by tag.