Fix leak of KVM GISC resources
Subsystems (5)
Anthony Krowiak
maintainerv6 cover Fix leak of KVM GISC resources
This is a continuation of the following patch:
[PATCH v5 9/9] s390/vfio-ap: Fix memory leak when queue removed from host
AP config
https://lore.kernel.org/linux-s390/[email protected]/
That patch was posted with a series of patches, but was no part of the pull request for the series due to some code review comments that needed to be addressed. The first 8/9 patches were part of a pull request to make the merge window which was about to close; so the referenced patch is posted on its own for review.
Anthony Krowiak (1):
s390/vfio-ap: fix KVM GISC and page leak when queue removed from host
config
drivers/s390/crypto/vfio_ap_ops.c | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-)
Signature · 1 line
2.53.0
Anthony Krowiak
maintainerPATCH v6 1/1 s390/vfio-ap: fix KVM GISC and page leak when queue removed from host config
Three related problems exist in the handling of KVM interrupt and page resources when a queue is removed from the host's AP configuration while assigned to a mediated device (mdev).
Problem 1:
~~~~~~~~~AP_RESPONSE_Q_NOT_AVAIL not handled in vfio_ap_mdev_reset_queue()
When the AP bus removes a queue device whose adapter or domain has been removed from the host's AP configuration, vfio_ap_mdev_remove_queue() is called. If the queue is still in the host's AP configuration at that point, it calls vfio_ap_mdev_reset_queue(), which issues a PQAP(ZAPQ). Since the adapter is already gone from the host configuration, ap_zapq() returns AP_RESPONSE_Q_NOT_AVAIL (0x01). This response code is not handled in vfio_ap_mdev_reset_queue()'s switch statement and falls through to the default case, which issues a WARN but does not call vfio_ap_free_aqic_resources(). As a result, if IRQ handling was enabled for the queue by the guest, the KVM GISC registration and the pinned guest page holding the notification indicator byte (NIB) are both leaked.
This is fixed by adding AP_RESPONSE_Q_NOT_AVAIL to the same case as AP_RESPONSE_DECONFIGURED and AP_RESPONSE_CHECKSTOPPED in vfio_ap_mdev_reset_queue(). Like those response codes, Q_NOT_AVAIL indicates the queue is not operational and no further reset attempts are possible; the correct action is to free the IRQ resources immediately.
Problem 2:
~~~~~~~~~
AP_RESPONSE_Q_NOT_AVAIL not handled in apq_status_check()
In vfio_ap_mdev_reset_queue(), there are four cases that indicate a queue reset has not yet completed, in which case apq_reset_check() is queued to a work queue to verify completion of the reset operation. This function uses the PQAP(TAPQ) function to get the queue's status and calls apq_status_check() to verify whether the reset has completed, failed or needs to be executed again. As described in Problem #1 above, apq_reset_check() does not specifically check for AP_RESPONSE_Q_NOT_AVAIL, thereby potentially leaking KVM GISC registration and the pinned guest page holding the NIB.
This is fixed by adding a case statement for AP_RESPONSE_Q_NOT_AVAIL to apq_status_check() and returning -ENODEV for that case. The caller, apq_reset_check() will then check for this return code and call vfio_ap_free_aqic_resources() to prevent the leak.
Problem 3:
~~~~~~~~~vfio_ap_free_aqic_resources() leaks saved_isc when kvm is NULL
vfio_ap_free_aqic_resources() guards the call tokvm_s390_gisc_unregister() with:
if (q->saved_isc != VFIO_AP_ISC_INVALID &&
!WARN_ON(!(q->matrix_mdev && q->matrix_mdev->kvm)))If matrix_mdev->kvm is NULL -- which can happen when vfio_ap_mdev_unset_kvm() has already run and cleared kvm before a subsequent cleanup path reaches this function -- the WARN_ON fires and the entire block is skipped. This leaves q->saved_isc set to a non-invalid value, creating a potential double-free on any subsequent call to this function.
When kvm is NULL the KVM guest is already torn down, so kvm_s390_gisc_unregister() need not and cannot be called; however, q->saved_isc must always be cleared. Fix this by separating the kvm_s390_gisc_unregister() call from the q->saved_isc reset. The WARN_ON now guards only the genuinely impossible case of matrix_mdev being NULL. A NULL kvm is handled gracefully by skipping only the unregister call, and q->saved_isc = VFIO_AP_ISC_INVALID is set unconditionally whenever saved_isc was not already invalid.
Additionally, add an else clause to the host-config check in vfio_ap_mdev_remove_queue() to call vfio_ap_free_aqic_resources() directly when the queue is not in the host's AP configuration. This serves as a backstop: when the AP bus fires the driver .remove callback after an adapter is removed from the host config, the queue is by definition no longer addressable, so vfio_ap_mdev_reset_queue() would always return Q_NOT_AVAIL. The else clause handles this case directly without the unnecessary ap_zapq() call, and ensures cleanup occurs even if kvm has already been set to NULL by a prior call to vfio_ap_mdev_unset_kvm().
--- drivers/s390/crypto/vfio_ap_ops.c | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-)
drivers/s390/crypto/vfio_ap_ops.c +14 -4
@@ -270,9 +270,9 @@ static void vfio_ap_free_aqic_resources(struct vfio_ap_queue *q) {if (!q)return ; -if (q->saved_isc !=VFIO_AP_ISC_INVALID && - !WARN_ON (!(q->matrix_mdev && q->matrix_mdev->kvm))) { -kvm_s390_gisc_unregister (q->matrix_mdev->kvm, q->saved_isc); +if (q->saved_isc !=VFIO_AP_ISC_INVALID ) { +if (!WARN_ON (!q->matrix_mdev) && q->matrix_mdev->kvm) +kvm_s390_gisc_unregister (q->matrix_mdev->kvm, q->saved_isc); q->saved_isc =VFIO_AP_ISC_INVALID ; }if (q->saved_iova && !WARN_ON (!q->matrix_mdev)) {
@@ -1935,6 +1935,8 @@ static int apq_status_check(int apqn, struct ap_queue_status *status)* a value indicating a reset needs to be performed again. */ return -EAGAIN ; +case AP_RESPONSE_Q_NOT_AVAIL : +return -ENODEV ;default : WARN(true, "failed to verify reset of queue %02x.%04x: TAPQ rc=%u\n",
@@ -1961,6 +1963,10 @@ static void apq_reset_check(struct work_struct *reset_work) ret =apq_status_check (q->apqn, &status);if (ret == -EIO )return ; +if (ret == -ENODEV ) { +vfio_ap_free_aqic_resources (q); +return ; + }if (ret == -EBUSY ) { pr_notice_ratelimited(WAIT_MSG, elapsed, AP_QID_CARD(q->apqn),
@@ -2004,6 +2010,7 @@ static void vfio_ap_mdev_reset_queue(struct vfio_ap_queue *q)break ;case AP_RESPONSE_DECONFIGURED :case AP_RESPONSE_CHECKSTOPPED : +case AP_RESPONSE_Q_NOT_AVAIL :vfio_ap_free_aqic_resources (q);break ;default :
@@ -2499,12 +2506,15 @@ void vfio_ap_mdev_remove_queue(struct ap_device *apdev)/* * If the queue is not in the host's AP configuration, then resetting * it will fail with response code 01, (APQN not valid); so, let's make -* sure it is in the host's config. +* sure it is in the host's config. If it is not, free the KVM GISC +* resources. */ if (test_bit_inv (apid, (unsigned long *)matrix_dev->info.apm) &&test_bit_inv (apqi, (unsigned long *)matrix_dev->info.aqm)) {vfio_ap_mdev_reset_queue (q);flush_work (&q->reset_work); + }else { +vfio_ap_free_aqic_resources (q); } done:
Signature · 1 line
2.53.0
[email protected]
AI reviewRe: [PATCH v6 1/1] s390/vfio-ap: fix KVM GISC and page leak when queue removed from host config
Sashiko has reviewed this patch and found no issues. It looks great!
Quoted 18 lines
Three related problems exist in the handling of KVM interrupt and page resources when a queue is removed from the host's AP configuration while assigned to a mediated device (mdev).
Problem 1:
~~~~~~~~~AP_RESPONSE_Q_NOT_AVAILnot handled invfio_ap_mdev_reset_queue()When the AP bus removes a queue device whose adapter or domain has been removed from the host's AP configuration,
vfio_ap_mdev_remove_queue()is called. If the queue is still in the host's AP configuration at that point, it callsvfio_ap_mdev_reset_queue(), which issues a PQAP(ZAPQ). Since the adapter is already gone from the host configuration,ap_zapq()returnsAP_RESPONSE_Q_NOT_AVAIL(0x01). This response code is not handled in [ ... ]vfio_ap_mdev_unset_kvm().
Quoted 3 lines
Fixes b9bd10c43456d ("s390/vfio-ap: do not reset queue removed from host config")Signed-off-by Anthony Krowiak <[email protected]>
Sashiko has reviewed this patch and found no issues. It looks great!
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Re: [PATCH v6 1/1] s390/vfio-ap: fix KVM GISC and page leak when queue removed from host config
In reply to Anthony KrowiakQuoted 4 lines · Anthony Krowiak
Three related problems exist in the handling of KVM interrupt and page resources when a queue is removed from the host's AP configuration while assigned to a mediated device (mdev).
...
Quoted 5 lines
Fixes b9bd10c43456d ("s390/vfio-ap: do not reset queue removed from host config")Signed-off-by Anthony Krowiak <[email protected]>Reviewed-by Matthew Rosato <[email protected]>
Changes LGTM, my review stands.
Thanks,
Matt
Re: [PATCH v6 1/1] s390/vfio-ap: fix KVM GISC and page leak when queue removed from host config
In reply to Anthony KrowiakQuoted 85 lines · On Tue, 18 Aug 2026 15:33:49 -0400 Anthony Krowiak <[email protected]>…
Three related problems exist in the handling of KVM interrupt and page resources when a queue is removed from the host's AP configuration while assigned to a mediated device (mdev).
Problem 1:
~~~~~~~~~AP_RESPONSE_Q_NOT_AVAILnot handled invfio_ap_mdev_reset_queue()When the AP bus removes a queue device whose adapter or domain has been removed from the host's AP configuration,
vfio_ap_mdev_remove_queue()is called. If the queue is still in the host's AP configuration at that point, it callsvfio_ap_mdev_reset_queue(), which issues a PQAP(ZAPQ). Since the adapter is already gone from the host configuration,ap_zapq()returnsAP_RESPONSE_Q_NOT_AVAIL(0x01). This response code is not handled in vfio_ap_mdev_reset_queue()'s switch statement and falls through to the default case, which issues a WARN but does not callvfio_ap_free_aqic_resources(). As a result, if IRQ handling was enabled for the queue by the guest, the KVM GISC registration and the pinned guest page holding the notification indicator byte (NIB) are both leaked.This is fixed by adding
AP_RESPONSE_Q_NOT_AVAILto the same case asAP_RESPONSE_DECONFIGUREDandAP_RESPONSE_CHECKSTOPPEDinvfio_ap_mdev_reset_queue(). Like those response codes,Q_NOT_AVAILindicates the queue is not operational and no further reset attempts are possible; the correct action is to free the IRQ resources immediately.Problem 2:
~~~~~~~~~
AP_RESPONSE_Q_NOT_AVAILnot handled inapq_status_check()In
vfio_ap_mdev_reset_queue(), there are four cases that indicate a queue reset has not yet completed, in which caseapq_reset_check()is queued to a work queue to verify completion of the reset operation. This function uses the PQAP(TAPQ) function to get the queue's status and callsapq_status_check()to verify whether the reset has completed, failed or needs to be executed again. As described in Problem #1 above,apq_reset_check()does not specifically check forAP_RESPONSE_Q_NOT_AVAIL, thereby potentially leaking KVM GISC registration and the pinned guest page holding the NIB.This is fixed by adding a case statement for
AP_RESPONSE_Q_NOT_AVAILtoapq_status_check()and returning-ENODEVfor that case. The caller,apq_reset_check()will then check for this return code and callvfio_ap_free_aqic_resources()to prevent the leak.Problem 3:
~~~~~~~~~vfio_ap_free_aqic_resources()leakssaved_iscwhen kvm is NULL
vfio_ap_free_aqic_resources()guards the call tokvm_s390_gisc_unregister()with:if (q->saved_isc != VFIO_AP_ISC_INVALID && !WARN_ON(!(q->matrix_mdev && q->matrix_mdev->kvm)))If
matrix_mdev->kvmis NULL -- which can happen whenvfio_ap_mdev_unset_kvm()has already run and cleared kvm before a subsequent cleanup path reaches this function -- theWARN_ONfires and the entire block is skipped. This leavesq->saved_iscset to a non-invalid value, creating a potential double-free on any subsequent call to this function.When kvm is NULL the KVM guest is already torn down, so
kvm_s390_gisc_unregister()need not and cannot be called; however,q->saved_iscmust always be cleared. Fix this by separating thekvm_s390_gisc_unregister()call from theq->saved_iscreset. TheWARN_ONnow guards only the genuinely impossible case ofmatrix_mdevbeing NULL. A NULL kvm is handled gracefully by skipping only the unregister call, andq->saved_isc=VFIO_AP_ISC_INVALIDis set unconditionally wheneversaved_iscwas not already invalid.Additionally, add an else clause to the host-config check in
vfio_ap_mdev_remove_queue()to callvfio_ap_free_aqic_resources()directly when the queue is not in the host's AP configuration. This serves as a backstop: when the AP bus fires the driver .remove callback after an adapter is removed from the host config, the queue is by definition no longer addressable, sovfio_ap_mdev_reset_queue()would always returnQ_NOT_AVAIL. The else clause handles this case directly without the unnecessaryap_zapq()call, and ensures cleanup occurs even if kvm has already been set to NULL by a prior call tovfio_ap_mdev_unset_kvm().
Quoted 4 lines
Fixes b9bd10c43456d ("s390/vfio-ap: do not reset queue removed from host config")Signed-off-by Anthony Krowiak <[email protected]>Reviewed-by Matthew Rosato <[email protected]>
Looks good to me functionally. Style-wise there are things we could talk about, but we don't have to :)
Christian Borntraeger
AppliedAm 18.08.26 um 21:33 schrieb Anthony Krowiak:
Quoted 12 lines
This is a continuation of the following patch:
[PATCH v5 9/9] s390/vfio-ap: Fix memory leak when queue removed from host
AP confighttps://lore.kernel.org/linux-s390/[email protected]/
That patch was posted with a series of patches, but was no part of the pull request for the series due to some code review comments that needed to be addressed. The first 8/9 patches were part of a pull request to make the merge window which was about to close; so the referenced patch is posted on its own for review.
Quoted 7 lines
Anthony Krowiak (1):
s390/vfio-ap: fix KVM GISC and page leak when queue removed from host
configdrivers/s390/crypto/vfio_ap_ops.c | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-)
Thanks, applied