Skip to contentSkip to conversation

Fix use-after-free and deadlock issues

PATCH Applied Xiaochen Shen 4 participants 20 messages Jan 8, 2020 – Feb 2, 2020 4 messages not in the archive

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover Fix use-after-free and deadlock issues

Fix several use-after-free and deadlock issues:
(1) Fix use-after-free when deleting resource groups.
(2) Fix use-after-free due to inaccurate refcount of rdtgroup.
(3) Fix a deadlock due to inaccurate active reference of kernfs node.
(4) Follow-up cleanup for the change in previous patch.

Xiaochen Shen (4):
x86/resctrl: Fix use-after-free when deleting resource groups x86/resctrl: Fix use-after-free due to inaccurate refcount of rdtgroup x86/resctrl: Fix a deadlock due to inaccurate active reference of
kernfs node
x86/resctrl: Clean up unused function parameter in mkdir path

 arch/x86/kernel/cpu/resctrl/rdtgroup.c | 48 ++++++++++++++++++----------------
 1 file changed, 25 insertions(+), 23 deletions(-)
Signature · 1 line
1.8.3.1
#1 lore raw

PATCH 1/4 x86/resctrl: Fix use-after-free when deleting resource groups

A resource group (rdtgrp) contains a reference count (rdtgrp->waitcount) that indicates how many waiters expect this rdtgrp to exist. Waiters could be waiting on rdtgroup_mutex or some work sitting on a task's workqueue for when the task returns from kernel mode or exits. The deletion of a rdtgrp is intended to have two phases: (1) while holding rdtgroup_mutex the necessary cleanup is done and rdtgrp->flags is set to RDT_DELETED, (2) after releasing the rdtgroup_mutex the rdtgrp structure is freed only if there are no waiters and it's flag is set to RDT_DELETED. Upon gaining access to rdtgroup_mutex or rdtgrp a waiter is required to check for the RDT_DELETED flag.

When unmounting resctrl file system or deleting ctrl_mon groups, all of the subdirectories are removed and the data structure of rdtgrp is forcibly freed without checking rdtgrp->waitcount. If at this point there was a waiter on rdtgrp then a use-after-free issue occurs when the waiter starts running accessing the rdtgrp structure it was waiting on.

See kfree() calls in [1], [2] and [3] in these two call paths in following scenarios:
(1) rdt_kill_sb() -> rmdir_all_sub() -> free_all_child_rdtgrp()
(2) rdtgroup_rmdir() -> rdtgroup_rmdir_ctrl() -> free_all_child_rdtgrp()

There are several scenarios that result in use-after-free issue in following:

Scenario 1:
-----------
In Thread 1, rdtgroup_tasks_write() adds a task_work callback move_myself(). If move_myself() is scheduled to execute after Thread 2 rdt_kill_sb() is finished, referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_tasks_write) Thread 2 (rdt_kill_sb)
------------------------------- ----------------------
rdtgroup_kn_lock_live
atomic_inc(&rdtgrp->waitcount)
mutex_lock
rdtgroup_move_task
__rdtgroup_move_task

    /*
     * Take an extra refcount, so rdtgrp cannot be freed
     * before the call back move_myself has been invoked
     */
    atomic_inc(&rdtgrp->waitcount)
    /* Callback move_myself will be scheduled for later */
    task_work_add(move_myself)

rdtgroup_kn_unlock
mutex_unlock
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

                                       mutex_lock
                                       rmdir_all_sub
                                         /*
                                          * sentry and rdtgrp are freed
                                          * without checking refcount
                                          */
                                         free_all_child_rdtgrp
                                           kfree(sentry)*[1]
                                         kfree(rdtgrp)*[2]
                                       mutex_unlock

/*
* Callback is scheduled to execute
* after rdt_kill_sb is finished
*/
move_myself
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [1] or [2].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)
kfree(rdtgrp)

Scenario 2:
-----------
In Thread 1, rdtgroup_tasks_write() adds a task_work callback move_myself(). If move_myself() is scheduled to execute after Thread 2 rdtgroup_rmdir() is finished, referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_tasks_write) Thread 2 (rdtgroup_rmdir)
------------------------------- -------------------------
rdtgroup_kn_lock_live
atomic_inc(&rdtgrp->waitcount)
mutex_lock
rdtgroup_move_task
__rdtgroup_move_task

    /*
     * Take an extra refcount, so rdtgrp cannot be freed
     * before the call back move_myself has been invoked
     */
    atomic_inc(&rdtgrp->waitcount)
    /* Callback move_myself will be scheduled for later */
    task_work_add(move_myself)

rdtgroup_kn_unlock
mutex_unlock
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

                                       rdtgroup_kn_lock_live
                                         atomic_inc(&rdtgrp->waitcount)
                                         mutex_lock
                                       rdtgroup_rmdir_ctrl
                                         free_all_child_rdtgrp
                                           /*
                                            * sentry is freed without
                                            * checking refcount
                                            */
                                           kfree(sentry)*[3]
                                         rdtgroup_ctrl_remove
                                           rdtgrp->flags = RDT_DELETED
                                       rdtgroup_kn_unlock
                                         mutex_unlock
                                         atomic_dec_and_test(
                                                     &rdtgrp->waitcount)
                                         && (flags & RDT_DELETED)
                                           kfree(rdtgrp)

/*
* Callback is scheduled to execute
* after rdt_kill_sb is finished
*/
move_myself
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [3].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)
kfree(rdtgrp)

If CONFIG_DEBUG_SLAB=y, Slab corruption on kmalloc-2k can be observed like following. Note that "0x6b" is POISON_FREE after kfree(). The corrupted bits "0x6a", "0x64" at offset 0x424 correspond to waitcount member of struct rdtgroup which was freed:

Slab corruption (Not tainted): kmalloc-2k start=ffff9504c5b0d000, len=2048 420: 6b 6b 6b 6b 6a 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkjkkkkkkkkkkk Single bit error detected. Probably bad RAM.
Run memtest86+ or a similar memory test tool.
Next obj: start=ffff9504c5b0d800, len=2048
000: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk 010: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk

Slab corruption (Not tainted): kmalloc-2k start=ffff9504c58ab800, len=2048 420: 6b 6b 6b 6b 64 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkdkkkkkkkkkkk Prev obj: start=ffff9504c58ab000, len=2048
000: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk 010: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk

Fix this by taking reference count (waitcount) of rdtgrp into account in the two call paths that currently do not do so. Instead of always freeing the resource group it will only be freed if there are no waiters on it, if there are waiters the resource group will have its flags set to RDT_DELETED. It will be left to the waiter to free the resource group when it starts running and finding that it was the last waiter and the resource group has been removed (rdtgrp->flags & RDT_DELETED) since.
(1) rdt_kill_sb() -> rmdir_all_sub() -> free_all_child_rdtgrp()
(2) rdtgroup_rmdir() -> rdtgroup_rmdir_ctrl() -> free_all_child_rdtgrp()

Fixes f3cbeacaa06e ("x86/intel_rdt/cqm: Add rmdir support")
Fixes 60cf5e101fd4 ("x86/intel_rdt: Add mkdir to resctrl file system")
---
 arch/x86/kernel/cpu/resctrl/rdtgroup.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
arch/x86/kernel/cpu/resctrl/rdtgroup.c +10 -2
index 205925d802d8..e1df86da48cf 100644
@@ -2209,7 +2209,11 @@ static void free_all_child_rdtgrp(struct rdtgroup *rdtgrp)
 	list_for_each_entry_safe(sentry, stmp, head, mon.crdtgrp_list) {
 		free_rmid(sentry->mon.rmid);
 		list_del(&sentry->mon.crdtgrp_list);
-		kfree(sentry);
+
+		if (atomic_read(&sentry->waitcount) != 0)
+			sentry->flags = RDT_DELETED;
+		else
+			kfree(sentry);
 	}
 }
 
@@ -2247,7 +2251,11 @@ static void rmdir_all_sub(void)
 
 		kernfs_remove(rdtgrp->kn);
 		list_del(&rdtgrp->rdtgroup_list);
-		kfree(rdtgrp);
+
+		if (atomic_read(&rdtgrp->waitcount) != 0)
+			rdtgrp->flags = RDT_DELETED;
+		else
+			kfree(rdtgrp);
 	}
 	/* Notify online CPUs to update per cpu storage and PQR_ASSOC MSR */
 	update_closid_rmid(cpu_online_mask, &rdtgroup_default);
Signature · 1 line
1.8.3.1

A message that is not in the archive (sent off-list or not yet received): [email protected]

#3 lore raw

Re: [PATCH 1/4] x86/resctrl: Fix use-after-free when deleting resource groups

Hi Sasha,

The backporting for v4.14.x and v4.19.x stable trees needs some manual tweaking due to difference code bases (since x86/resctrl rename/re-arrange in v5.0 upstream kernel).

I will work on the backporting once this patch is merged into upstream.

Thank you.

Quoted 21 lines · Sasha Levin

Hi,

[This is an automated email]

This commit has been processed because it contains a "Fixes:" tag, fixing commit: f3cbeacaa06e ("x86/intel_rdt/cqm: Add rmdir support").

The bot has tested the following trees: v5.4.8, v4.19.93, v4.14.162.

v5.4.8: Build OK!
v4.19.93: Failed to apply! Possible dependencies:
Unable to calculate

v4.14.162: Failed to apply! Possible dependencies:
Unable to calculate

NOTE: The patch will not be queued to stable trees until it is upstream.

How should we proceed with this patch?

Signature · 2 lines
Best regards,
Xiaochen
merge bot Applied
#4 lore raw

[tip: x86/urgent] x86/resctrl: Fix use-after-free when deleting resource groups

In reply to Xiaochen Shen
The following commit has been merged into the x86/urgent branch of tip:

The following commit has been merged into the x86/urgent branch of tip:

Commit-ID:     b8511ccc75c033f6d54188ea4df7bf1e85778740
Gitweb:        https://git.kernel.org/tip/b8511ccc75c033f6d54188ea4df7bf1e85778740
Author:        Xiaochen Shen <[email protected]>
AuthorDate:    Thu, 09 Jan 2020 00:28:03 +08:00
Committer:     Borislav Petkov <[email protected]>
CommitterDate: Mon, 20 Jan 2020 16:45:43 +01:00

x86/resctrl: Fix use-after-free when deleting resource groups

A resource group (rdtgrp) contains a reference count (rdtgrp->waitcount) that indicates how many waiters expect this rdtgrp to exist. Waiters could be waiting on rdtgroup_mutex or some work sitting on a task's workqueue for when the task returns from kernel mode or exits.

The deletion of a rdtgrp is intended to have two phases:

(1) while holding rdtgroup_mutex the necessary cleanup is done and rdtgrp->flags is set to RDT_DELETED,

(2) after releasing the rdtgroup_mutex, the rdtgrp structure is freed only if there are no waiters and its flag is set to RDT_DELETED. Upon gaining access to rdtgroup_mutex or rdtgrp, a waiter is required to check for the RDT_DELETED flag.

When unmounting the resctrl file system or deleting ctrl_mon groups, all of the subdirectories are removed and the data structure of rdtgrp is forcibly freed without checking rdtgrp->waitcount. If at this point there was a waiter on rdtgrp then a use-after-free issue occurs when the waiter starts running and accesses the rdtgrp structure it was waiting on.

See kfree() calls in [1], [2] and [3] in these two call paths in following scenarios:
(1) rdt_kill_sb() -> rmdir_all_sub() -> free_all_child_rdtgrp()
(2) rdtgroup_rmdir() -> rdtgroup_rmdir_ctrl() -> free_all_child_rdtgrp()

There are several scenarios that result in use-after-free issue in following:

Scenario 1:
-----------
In Thread 1, rdtgroup_tasks_write() adds a task_work callback move_myself(). If move_myself() is scheduled to execute after Thread 2 rdt_kill_sb() is finished, referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_tasks_write) Thread 2 (rdt_kill_sb)
------------------------------- ----------------------
rdtgroup_kn_lock_live
atomic_inc(&rdtgrp->waitcount)
mutex_lock
rdtgroup_move_task
__rdtgroup_move_task

    /*
     * Take an extra refcount, so rdtgrp cannot be freed
     * before the call back move_myself has been invoked
     */
    atomic_inc(&rdtgrp->waitcount)
    /* Callback move_myself will be scheduled for later */
    task_work_add(move_myself)

rdtgroup_kn_unlock
mutex_unlock
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

                                       mutex_lock
                                       rmdir_all_sub
                                         /*
                                          * sentry and rdtgrp are freed
                                          * without checking refcount
                                          */
                                         free_all_child_rdtgrp
                                           kfree(sentry)*[1]
                                         kfree(rdtgrp)*[2]
                                       mutex_unlock

/*
* Callback is scheduled to execute
* after rdt_kill_sb is finished
*/
move_myself
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [1] or [2].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)
kfree(rdtgrp)

Scenario 2:
-----------
In Thread 1, rdtgroup_tasks_write() adds a task_work callback move_myself(). If move_myself() is scheduled to execute after Thread 2 rdtgroup_rmdir() is finished, referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_tasks_write) Thread 2 (rdtgroup_rmdir)
------------------------------- -------------------------
rdtgroup_kn_lock_live
atomic_inc(&rdtgrp->waitcount)
mutex_lock
rdtgroup_move_task
__rdtgroup_move_task

    /*
     * Take an extra refcount, so rdtgrp cannot be freed
     * before the call back move_myself has been invoked
     */
    atomic_inc(&rdtgrp->waitcount)
    /* Callback move_myself will be scheduled for later */
    task_work_add(move_myself)

rdtgroup_kn_unlock
mutex_unlock
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

                                       rdtgroup_kn_lock_live
                                         atomic_inc(&rdtgrp->waitcount)
                                         mutex_lock
                                       rdtgroup_rmdir_ctrl
                                         free_all_child_rdtgrp
                                           /*
                                            * sentry is freed without
                                            * checking refcount
                                            */
                                           kfree(sentry)*[3]
                                         rdtgroup_ctrl_remove
                                           rdtgrp->flags = RDT_DELETED
                                       rdtgroup_kn_unlock
                                         mutex_unlock
                                         atomic_dec_and_test(
                                                     &rdtgrp->waitcount)
                                         && (flags & RDT_DELETED)
                                           kfree(rdtgrp)

/*
* Callback is scheduled to execute
* after rdt_kill_sb is finished
*/
move_myself
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [3].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)
kfree(rdtgrp)

If CONFIG_DEBUG_SLAB=y, Slab corruption on kmalloc-2k can be observed like following. Note that "0x6b" is POISON_FREE after kfree(). The corrupted bits "0x6a", "0x64" at offset 0x424 correspond to waitcount member of struct rdtgroup which was freed:

Slab corruption (Not tainted): kmalloc-2k start=ffff9504c5b0d000, len=2048 420: 6b 6b 6b 6b 6a 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkjkkkkkkkkkkk Single bit error detected. Probably bad RAM.
Run memtest86+ or a similar memory test tool.
Next obj: start=ffff9504c5b0d800, len=2048
000: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk 010: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk

Slab corruption (Not tainted): kmalloc-2k start=ffff9504c58ab800, len=2048 420: 6b 6b 6b 6b 64 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkdkkkkkkkkkkk Prev obj: start=ffff9504c58ab000, len=2048
000: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk 010: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk

Fix this by taking reference count (waitcount) of rdtgrp into account in the two call paths that currently do not do so. Instead of always freeing the resource group it will only be freed if there are no waiters on it. If there are waiters, the resource group will have its flags set to RDT_DELETED.

It will be left to the waiter to free the resource group when it starts running and finding that it was the last waiter and the resource group has been removed (rdtgrp->flags & RDT_DELETED) since. (1) rdt_kill_sb() -> rmdir_all_sub() -> free_all_child_rdtgrp() (2) rdtgroup_rmdir() -> rdtgroup_rmdir_ctrl() -> free_all_child_rdtgrp()

Fixes f3cbeacaa06e ("x86/intel_rdt/cqm: Add rmdir support")
Fixes 60cf5e101fd4 ("x86/intel_rdt: Add mkdir to resctrl file system")
---
 arch/x86/kernel/cpu/resctrl/rdtgroup.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
arch/x86/kernel/cpu/resctrl/rdtgroup.c +10 -2
index dac7209..23904ab 100644
@@ -2205,7 +2205,11 @@ static void free_all_child_rdtgrp(struct rdtgroup *rdtgrp)
 	list_for_each_entry_safe(sentry, stmp, head, mon.crdtgrp_list) {
 		free_rmid(sentry->mon.rmid);
 		list_del(&sentry->mon.crdtgrp_list);
-		kfree(sentry);
+
+		if (atomic_read(&sentry->waitcount) != 0)
+			sentry->flags = RDT_DELETED;
+		else
+			kfree(sentry);
 	}
 }
 
@@ -2243,7 +2247,11 @@ static void rmdir_all_sub(void)
 
 		kernfs_remove(rdtgrp->kn);
 		list_del(&rdtgrp->rdtgroup_list);
-		kfree(rdtgrp);
+
+		if (atomic_read(&rdtgrp->waitcount) != 0)
+			rdtgrp->flags = RDT_DELETED;
+		else
+			kfree(rdtgrp);
 	}
 	/* Notify online CPUs to update per cpu storage and PQR_ASSOC MSR */
 	update_closid_rmid(cpu_online_mask, &rdtgroup_default);
#5 lore raw

PATCH 2/4 x86/resctrl: Fix use-after-free due to inaccurate refcount of rdtgroup

In reply to Xiaochen Shen

There is a race condition in following scenario which results in use-after-free issue when reading a monitoring file and deleting the parent ctrl_mon group concurrently:

Thread 1 calls atomic_inc() to take refcount of rdtgrp and then calls kernfs_break_active_protection() to drop the active reference of kernfs node in rdtgroup_kn_lock_live().

In Thread 2, kernfs_remove() is a blocking routine. It waits on all sub kernfs nodes to drop the active reference when removing all subtree kernfs nodes recursively. Thread 2 could block on kernfs_remove() until Thread 1 calls kernfs_break_active_protection(). Only after kernfs_remove() completes the refcount of rdtgrp could be trusted.

Before Thread 1 calls atomic_inc() and kernfs_break_active_protection(), Thread 2 could call kfree() when the refcount of rdtgrp (sentry) is 0 instead of 1 due to the race.

In Thread 1, in rdtgroup_kn_unlock(), referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_mondata_show) Thread 2 (rdtgroup_rmdir) -------------------------------- ------------------------- rdtgroup_kn_lock_live
/*
* kn active protection until
* kernfs_break_active_protection(kn)
*/
rdtgrp = kernfs_to_rdtgroup(kn)

                                  rdtgroup_kn_lock_live
                                    atomic_inc(&rdtgrp->waitcount)
                                    mutex_lock
                                  rdtgroup_rmdir_ctrl
                                    free_all_child_rdtgrp
                                      /*
                                       * sentry->waitcount should be 1
                                       * but is 0 now due to the race.
                                       */
                                      kfree(sentry)*[1]
  /*
   * Only after kernfs_remove()
   * completes, the refcount of
   * rdtgrp could be trusted.
   */
  atomic_inc(&rdtgrp->waitcount)
  /* kn->active-- */
  kernfs_break_active_protection(kn)
                                    rdtgroup_ctrl_remove
                                      rdtgrp->flags = RDT_DELETED
                                      /*
                                       * Blocking routine, wait for
                                       * all sub kernfs nodes to drop
                                       * active reference in
                                       * kernfs_break_active_protection.
                                       */
                                      kernfs_remove(rdtgrp->kn)
                                  rdtgroup_kn_unlock
                                    mutex_unlock
                                    atomic_dec_and_test(
                                                &rdtgrp->waitcount)
                                    && (flags & RDT_DELETED)
                                      kernfs_unbreak_active_protection(kn)
                                      kfree(rdtgrp)

mutex_lock
mon_event_read
rdtgroup_kn_unlock
mutex_unlock
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [1].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

    /* kn->active++ */
    kernfs_unbreak_active_protection(kn)
    kfree(rdtgrp)

Fix it by moving free_all_child_rdtgrp() to after kernfs_remove() in rdtgroup_rmdir_ctrl() to ensure it has the accurate refcount of rdtgrp.

Fixes f3cbeacaa06e ("x86/intel_rdt/cqm: Add rmdir support")
---
 arch/x86/kernel/cpu/resctrl/rdtgroup.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/x86/kernel/cpu/resctrl/rdtgroup.c +2 -2
index e1df86da48cf..bb911c8f7cac 100644
@@ -2964,13 +2964,13 @@ static int rdtgroup_rmdir_ctrl(struct kernfs_node *kn, struct rdtgroup *rdtgrp,
 	closid_free(rdtgrp->closid);
 	free_rmid(rdtgrp->mon.rmid);
 
+	rdtgroup_ctrl_remove(kn, rdtgrp);
+
 	/*
 	 * Free all the child monitor group rmids.
 	 */
 	free_all_child_rdtgrp(rdtgrp);
 
-	rdtgroup_ctrl_remove(kn, rdtgrp);
-
 	return 0;
 }
 
Signature · 1 line
1.8.3.1

A message that is not in the archive (sent off-list or not yet received): [email protected]

#7 lore raw

Re: [PATCH 2/4] x86/resctrl: Fix use-after-free due to inaccurate refcount of rdtgroup

Hi Sasha,

The backporting for v4.14.x stable tree needs some manual tweaking due to difference code bases:
1. x86/resctrl rename/re-arrange in v5.0 upstream kernel.
2. Code re-arrange in upstream commit 17eafd076291 ("x86/intel_rdt: Split resource group removal in two").

The backporting for v4.19.x stable tree needs some manual tweaking due to difference code bases: x86/resctrl rename/re-arrange in v5.0 upstream kernel.

I will work on the backporting once this patch is merged into upstream.

Thank you.

Quoted 21 lines · Sasha Levin

Hi,

[This is an automated email]

This commit has been processed because it contains a "Fixes:" tag, fixing commit: f3cbeacaa06e ("x86/intel_rdt/cqm: Add rmdir support").

The bot has tested the following trees: v5.4.8, v4.19.93, v4.14.162.

v5.4.8: Build OK!
v4.19.93: Failed to apply! Possible dependencies:
Unable to calculate

v4.14.162: Failed to apply! Possible dependencies:

     17eafd076291 ("x86/intel_rdt: Split resource group removal in two")

NOTE: The patch will not be queued to stable trees until it is upstream.

How should we proceed with this patch?

Signature · 2 lines
Best regards,
Xiaochen
merge bot Applied
#8 lore raw

[tip: x86/urgent] x86/resctrl: Fix use-after-free due to inaccurate refcount of rdtgroup

In reply to Xiaochen Shen
The following commit has been merged into the x86/urgent branch of tip:

The following commit has been merged into the x86/urgent branch of tip:

Commit-ID:     074fadee59ee7a9d2b216e9854bd4efb5dad679f
Gitweb:        https://git.kernel.org/tip/074fadee59ee7a9d2b216e9854bd4efb5dad679f
Author:        Xiaochen Shen <[email protected]>
AuthorDate:    Thu, 09 Jan 2020 00:28:04 +08:00
Committer:     Borislav Petkov <[email protected]>
CommitterDate: Mon, 20 Jan 2020 16:56:11 +01:00

x86/resctrl: Fix use-after-free due to inaccurate refcount of rdtgroup

There is a race condition in the following scenario which results in an use-after-free issue when reading a monitoring file and deleting the parent ctrl_mon group concurrently:

Thread 1 calls atomic_inc() to take refcount of rdtgrp and then calls kernfs_break_active_protection() to drop the active reference of kernfs node in rdtgroup_kn_lock_live().

In Thread 2, kernfs_remove() is a blocking routine. It waits on all sub kernfs nodes to drop the active reference when removing all subtree kernfs nodes recursively. Thread 2 could block on kernfs_remove() until Thread 1 calls kernfs_break_active_protection(). Only after kernfs_remove() completes the refcount of rdtgrp could be trusted.

Before Thread 1 calls atomic_inc() and kernfs_break_active_protection(), Thread 2 could call kfree() when the refcount of rdtgrp (sentry) is 0 instead of 1 due to the race.

In Thread 1, in rdtgroup_kn_unlock(), referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_mondata_show) Thread 2 (rdtgroup_rmdir) -------------------------------- ------------------------- rdtgroup_kn_lock_live
/*
* kn active protection until
* kernfs_break_active_protection(kn)
*/
rdtgrp = kernfs_to_rdtgroup(kn)

                                  rdtgroup_kn_lock_live
                                    atomic_inc(&rdtgrp->waitcount)
                                    mutex_lock
                                  rdtgroup_rmdir_ctrl
                                    free_all_child_rdtgrp
                                      /*
                                       * sentry->waitcount should be 1
                                       * but is 0 now due to the race.
                                       */
                                      kfree(sentry)*[1]
  /*
   * Only after kernfs_remove()
   * completes, the refcount of
   * rdtgrp could be trusted.
   */
  atomic_inc(&rdtgrp->waitcount)
  /* kn->active-- */
  kernfs_break_active_protection(kn)
                                    rdtgroup_ctrl_remove
                                      rdtgrp->flags = RDT_DELETED
                                      /*
                                       * Blocking routine, wait for
                                       * all sub kernfs nodes to drop
                                       * active reference in
                                       * kernfs_break_active_protection.
                                       */
                                      kernfs_remove(rdtgrp->kn)
                                  rdtgroup_kn_unlock
                                    mutex_unlock
                                    atomic_dec_and_test(
                                                &rdtgrp->waitcount)
                                    && (flags & RDT_DELETED)
                                      kernfs_unbreak_active_protection(kn)
                                      kfree(rdtgrp)

mutex_lock
mon_event_read
rdtgroup_kn_unlock
mutex_unlock
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [1].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

    /* kn->active++ */
    kernfs_unbreak_active_protection(kn)
    kfree(rdtgrp)

Fix it by moving free_all_child_rdtgrp() to after kernfs_remove() in rdtgroup_rmdir_ctrl() to ensure it has the accurate refcount of rdtgrp.

---
 arch/x86/kernel/cpu/resctrl/rdtgroup.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/x86/kernel/cpu/resctrl/rdtgroup.c +2 -2
index 23904ab..caab397 100644
@@ -2960,13 +2960,13 @@ static int rdtgroup_rmdir_ctrl(struct kernfs_node *kn, struct rdtgroup *rdtgrp,
 	closid_free(rdtgrp->closid);
 	free_rmid(rdtgrp->mon.rmid);
 
+	rdtgroup_ctrl_remove(kn, rdtgrp);
+
 	/*
 	 * Free all the child monitor group rmids.
 	 */
 	free_all_child_rdtgrp(rdtgrp);
 
-	rdtgroup_ctrl_remove(kn, rdtgrp);
-
 	return 0;
 }
 
#9 lore raw

PATCH 3/4 x86/resctrl: Fix a deadlock due to inaccurate active reference of kernfs node

In reply to Xiaochen Shen

There is a race condition which results in a deadlock when rmdir and mkdir concurrently:

$ ls /sys/fs/resctrl/c1/mon_groups/m1/
cpus  cpus_list  mon_data  tasks

Thread 1: rmdir /sys/fs/resctrl/c1
Thread 2: mkdir /sys/fs/resctrl/c1/mon_groups/m1

3 locks held by mkdir/48649:

 #0:  (sb_writers#17){.+.+}, at: [<ffffffffb4ca2aa0>] mnt_want_write+0x20/0x50
 #1:  (&type->i_mutex_dir_key#8/1){+.+.}, at: [<ffffffffb4c8c13b>] filename_create+0x7b/0x170
 #2:  (rdtgroup_mutex){+.+.}, at: [<ffffffffb4a4389d>] rdtgroup_kn_lock_live+0x3d/0x70

4 locks held by rmdir/48652:

 #0:  (sb_writers#17){.+.+}, at: [<ffffffffb4ca2aa0>] mnt_want_write+0x20/0x50
 #1:  (&type->i_mutex_dir_key#8/1){+.+.}, at: [<ffffffffb4c8c3cf>] do_rmdir+0x13f/0x1e0
 #2:  (&type->i_mutex_dir_key#8){++++}, at: [<ffffffffb4c86d5d>] vfs_rmdir+0x4d/0x120
 #3:  (rdtgroup_mutex){+.+.}, at: [<ffffffffb4a4389d>] rdtgroup_kn_lock_live+0x3d/0x70

Thread 1 is deleting control group "c1". Holding rdtgroup_mutex, kernfs_remove() removes all kernfs nodes under directory "c1" recursively, then waits for sub kernfs node "mon_groups" to drop active reference.

Thread 2 is trying to create a subdirectory "m1" in the "mon_groups" directory. The wrapper kernfs_iop_mkdir() takes an active reference to the "mon_groups" directory but the code drops the active reference to the parent directory "c1" instead.

As a result, Thread 1 is blocked on waiting for active reference to drop and never release rdtgroup_mutex, while Thread 2 is also blocked on trying to get rdtgroup_mutex.

Thread 1 (rdtgroup_rmdir)   Thread 2 (rdtgroup_mkdir)
(rmdir /sys/fs/resctrl/c1)  (mkdir /sys/fs/resctrl/c1/mon_groups/m1)
-------------------------   -------------------------

kernfs_iop_mkdir
/*
* kn: "m1", parent_kn: "mon_groups",
* prgrp_kn: parent_kn->parent: "c1", *
* "mon_groups", parent_kn->active++: 1 */
kernfs_get_active(parent_kn)
kernfs_iop_rmdir
/* "c1", kn->active++ */
kernfs_get_active(kn)

rdtgroup_kn_lock_live

    atomic_inc(&rdtgrp->waitcount)
    /* "c1", kn->active-- */
    kernfs_break_active_protection(kn)
    mutex_lock

rdtgroup_rmdir_ctrl

    free_all_child_rdtgrp
      sentry->flags = RDT_DELETED

    rdtgroup_ctrl_remove
      rdtgrp->flags = RDT_DELETED
      kernfs_get(kn)
      kernfs_remove(rdtgrp->kn)
        __kernfs_remove
          /* "mon_groups", sub_kn */
          atomic_add(KN_DEACTIVATED_BIAS, &sub_kn->active)
          kernfs_drain(sub_kn)
            /*
             * sub_kn->active == KN_DEACTIVATED_BIAS + 1,
             * waiting on sub_kn->active to drop, but it
             * never drops in Thread 2 which is blocked
             * on getting rdtgroup_mutex.
             */

Thread 1 hangs here ---->
wait_event(sub_kn->active == KN_DEACTIVATED_BIAS)
...
rdtgroup_mkdir
rdtgroup_mkdir_mon(parent_kn, prgrp_kn)
mkdir_rdt_prepare(parent_kn, prgrp_kn)
rdtgroup_kn_lock_live(prgrp_kn)
atomic_inc(&rdtgrp->waitcount) /*
* "c1", prgrp_kn->active-- *
* The active reference on "c1" is
* dropped, but not matching the
* actual active reference taken
* on "mon_groups", thus causing
* Thread 1 to wait forever while
* holding rdtgroup_mutex. */
kernfs_break_active_protection(
prgrp_kn)
/*
* Trying to get rdtgroup_mutex
* which is held by Thread 1. */
Thread 2 hangs here ----> mutex_lock

                                      ...

The problem is that the creation of a subdirectory in the "mon_groups" directory incorrectly releases the active protection of its parent directory instead of itself before it starts waiting for rdtgroup_mutex. This is triggered by the rdtgroup_mkdir() flow calling rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() with kernfs node of the parent control group ("c1") as argument. It should be called with kernfs node "mon_groups" instead. What is currently missing is that the kn->priv of "mon_groups" is NULL instead of pointing to the rdtgrp.

Fix it by pointing kn->priv to rdtgrp when "mon_groups" is created. Then it could be passed to rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() instead. And then it operates on the same rdtgroup structure but handles the active reference of kernfs node "mon_groups" to prevent deadlock. The same changes are also made to the "mon_data" directories.

This patch results in some unused function parameters that will be cleaned up in follow-up patch, this patch focuses on the fix only in support of backport efforts.

Fixes c7d9aac61311 ("x86/intel_rdt/cqm: Add mkdir support for RDT monitoring")
---
 arch/x86/kernel/cpu/resctrl/rdtgroup.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
arch/x86/kernel/cpu/resctrl/rdtgroup.c +8 -8
index bb911c8f7cac..2c79fcd83271 100644
@@ -1974,7 +1974,7 @@ static int rdt_get_tree(struct fs_context *fc)
 
 	if (rdt_mon_capable) {
 		ret = mongroup_create_dir(rdtgroup_default.kn,
-					  NULL, "mon_groups",
+					  &rdtgroup_default, "mon_groups",
 					  &kn_mongrp);
 		if (ret < 0)
 			goto out_info;
@@ -2458,7 +2458,7 @@ static int mkdir_mondata_all(struct kernfs_node *parent_kn,
 	/*
 	 * Create the mon_data directory first.
 	 */
-	ret = mongroup_create_dir(parent_kn, NULL, "mon_data", &kn);
+	ret = mongroup_create_dir(parent_kn, prgrp, "mon_data", &kn);
 	if (ret)
 		return ret;
 
@@ -2657,7 +2657,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 	uint files = 0;
 	int ret;
 
-	prdtgrp = rdtgroup_kn_lock_live(prgrp_kn);
+	prdtgrp = rdtgroup_kn_lock_live(parent_kn);
 	if (!prdtgrp) {
 		ret = -ENODEV;
 		goto out_unlock;
@@ -2730,7 +2730,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 	kernfs_activate(kn);
 
 	/*
-	 * The caller unlocks the prgrp_kn upon success.
+	 * The caller unlocks the parent_kn upon success.
 	 */
 	return 0;
 
@@ -2741,7 +2741,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 out_free_rgrp:
 	kfree(rdtgrp);
 out_unlock:
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
@@ -2779,7 +2779,7 @@ static int rdtgroup_mkdir_mon(struct kernfs_node *parent_kn,
 	 */
 	list_add_tail(&rdtgrp->mon.crdtgrp_list, &prgrp->mon.crdtgrp_list);
 
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
@@ -2822,7 +2822,7 @@ static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
 		 * Create an empty mon_groups directory to hold the subset
 		 * of tasks and cpus to monitor.
 		 */
-		ret = mongroup_create_dir(kn, NULL, "mon_groups", NULL);
+		ret = mongroup_create_dir(kn, rdtgrp, "mon_groups", NULL);
 		if (ret) {
 			rdt_last_cmd_puts("kernfs subdir error\n");
 			goto out_del_list;
@@ -2838,7 +2838,7 @@ static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
 out_common_fail:
 	mkdir_rdt_prepare_clean(rdtgrp);
 out_unlock:
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
Signature · 1 line
1.8.3.1

A message that is not in the archive (sent off-list or not yet received): [email protected]

#11 lore raw

Re: [PATCH 3/4] x86/resctrl: Fix a deadlock due to inaccurate active reference of kernfs node

Hi Sasha,

The backporting for v4.14.x stable tree needs some manual tweaking due to difference code bases:
1. x86/resctrl rename/re-arrange in v5.0 upstream kernel.
2. Code change in upstream commit cfd0f34e4cd5 ("x86/intel_rdt: Add diagnostics when making directories").

The backporting for v4.19.x stable tree needs some manual tweaking due to difference code bases: x86/resctrl rename/re-arrange in v5.0 upstream kernel.

I will work on the backporting once this patch is merged into upstream.

Thank you.

Quoted 21 lines · Sasha Levin

Hi,

[This is an automated email]

This commit has been processed because it contains a "Fixes:" tag,
fixing commit: c7d9aac61311 ("x86/intel_rdt/cqm: Add mkdir support for RDT monitoring").

The bot has tested the following trees: v5.4.8, v4.19.93, v4.14.162.

v5.4.8: Build OK!
v4.19.93: Failed to apply! Possible dependencies:
Unable to calculate

v4.14.162: Failed to apply! Possible dependencies:
cfd0f34e4cd5 ("x86/intel_rdt: Add diagnostics when making directories")

NOTE: The patch will not be queued to stable trees until it is upstream.

How should we proceed with this patch?

Signature · 2 lines
Best regards,
Xiaochen
merge bot Applied
#12 lore raw

[tip: x86/urgent] x86/resctrl: Fix a deadlock due to inaccurate reference

In reply to Xiaochen Shen
The following commit has been merged into the x86/urgent branch of tip:

The following commit has been merged into the x86/urgent branch of tip:

Commit-ID:     334b0f4e9b1b4a1d475f803419d202f6c5e4d18e
Gitweb:        https://git.kernel.org/tip/334b0f4e9b1b4a1d475f803419d202f6c5e4d18e
Author:        Xiaochen Shen <[email protected]>
AuthorDate:    Thu, 09 Jan 2020 00:28:05 +08:00
Committer:     Borislav Petkov <[email protected]>
CommitterDate: Mon, 20 Jan 2020 16:57:53 +01:00

x86/resctrl: Fix a deadlock due to inaccurate reference

There is a race condition which results in a deadlock when rmdir and mkdir execute concurrently:

$ ls /sys/fs/resctrl/c1/mon_groups/m1/
cpus  cpus_list  mon_data  tasks

Thread 1: rmdir /sys/fs/resctrl/c1
Thread 2: mkdir /sys/fs/resctrl/c1/mon_groups/m1

3 locks held by mkdir/48649:

 #0:  (sb_writers#17){.+.+}, at: [<ffffffffb4ca2aa0>] mnt_want_write+0x20/0x50
 #1:  (&type->i_mutex_dir_key#8/1){+.+.}, at: [<ffffffffb4c8c13b>] filename_create+0x7b/0x170
 #2:  (rdtgroup_mutex){+.+.}, at: [<ffffffffb4a4389d>] rdtgroup_kn_lock_live+0x3d/0x70

4 locks held by rmdir/48652:

 #0:  (sb_writers#17){.+.+}, at: [<ffffffffb4ca2aa0>] mnt_want_write+0x20/0x50
 #1:  (&type->i_mutex_dir_key#8/1){+.+.}, at: [<ffffffffb4c8c3cf>] do_rmdir+0x13f/0x1e0
 #2:  (&type->i_mutex_dir_key#8){++++}, at: [<ffffffffb4c86d5d>] vfs_rmdir+0x4d/0x120
 #3:  (rdtgroup_mutex){+.+.}, at: [<ffffffffb4a4389d>] rdtgroup_kn_lock_live+0x3d/0x70

Thread 1 is deleting control group "c1". Holding rdtgroup_mutex, kernfs_remove() removes all kernfs nodes under directory "c1" recursively, then waits for sub kernfs node "mon_groups" to drop active reference.

Thread 2 is trying to create a subdirectory "m1" in the "mon_groups" directory. The wrapper kernfs_iop_mkdir() takes an active reference to the "mon_groups" directory but the code drops the active reference to the parent directory "c1" instead.

As a result, Thread 1 is blocked on waiting for active reference to drop and never release rdtgroup_mutex, while Thread 2 is also blocked on trying to get rdtgroup_mutex.

Thread 1 (rdtgroup_rmdir)   Thread 2 (rdtgroup_mkdir)
(rmdir /sys/fs/resctrl/c1)  (mkdir /sys/fs/resctrl/c1/mon_groups/m1)
-------------------------   -------------------------

kernfs_iop_mkdir
/*
* kn: "m1", parent_kn: "mon_groups",
* prgrp_kn: parent_kn->parent: "c1", *
* "mon_groups", parent_kn->active++: 1 */
kernfs_get_active(parent_kn)
kernfs_iop_rmdir
/* "c1", kn->active++ */
kernfs_get_active(kn)

rdtgroup_kn_lock_live

    atomic_inc(&rdtgrp->waitcount)
    /* "c1", kn->active-- */
    kernfs_break_active_protection(kn)
    mutex_lock

rdtgroup_rmdir_ctrl

    free_all_child_rdtgrp
      sentry->flags = RDT_DELETED

    rdtgroup_ctrl_remove
      rdtgrp->flags = RDT_DELETED
      kernfs_get(kn)
      kernfs_remove(rdtgrp->kn)
        __kernfs_remove
          /* "mon_groups", sub_kn */
          atomic_add(KN_DEACTIVATED_BIAS, &sub_kn->active)
          kernfs_drain(sub_kn)
            /*
             * sub_kn->active == KN_DEACTIVATED_BIAS + 1,
             * waiting on sub_kn->active to drop, but it
             * never drops in Thread 2 which is blocked
             * on getting rdtgroup_mutex.
             */

Thread 1 hangs here ---->
wait_event(sub_kn->active == KN_DEACTIVATED_BIAS)
...
rdtgroup_mkdir
rdtgroup_mkdir_mon(parent_kn, prgrp_kn)
mkdir_rdt_prepare(parent_kn, prgrp_kn)
rdtgroup_kn_lock_live(prgrp_kn)
atomic_inc(&rdtgrp->waitcount) /*
* "c1", prgrp_kn->active-- *
* The active reference on "c1" is
* dropped, but not matching the
* actual active reference taken
* on "mon_groups", thus causing
* Thread 1 to wait forever while
* holding rdtgroup_mutex. */
kernfs_break_active_protection(
prgrp_kn)
/*
* Trying to get rdtgroup_mutex
* which is held by Thread 1. */
Thread 2 hangs here ----> mutex_lock

                                      ...

The problem is that the creation of a subdirectory in the "mon_groups" directory incorrectly releases the active protection of its parent directory instead of itself before it starts waiting for rdtgroup_mutex. This is triggered by the rdtgroup_mkdir() flow calling rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() with kernfs node of the parent control group ("c1") as argument. It should be called with kernfs node "mon_groups" instead. What is currently missing is that the kn->priv of "mon_groups" is NULL instead of pointing to the rdtgrp.

Fix it by pointing kn->priv to rdtgrp when "mon_groups" is created. Then it could be passed to rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() instead. And then it operates on the same rdtgroup structure but handles the active reference of kernfs node "mon_groups" to prevent deadlock. The same changes are also made to the "mon_data" directories.

This results in some unused function parameters that will be cleaned up in follow-up patch as the focus here is on the fix only in support of backporting efforts.

---
 arch/x86/kernel/cpu/resctrl/rdtgroup.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
arch/x86/kernel/cpu/resctrl/rdtgroup.c +8 -8
index caab397..954fd04 100644
@@ -1970,7 +1970,7 @@ static int rdt_get_tree(struct fs_context *fc)
 
 	if (rdt_mon_capable) {
 		ret = mongroup_create_dir(rdtgroup_default.kn,
-					  NULL, "mon_groups",
+					  &rdtgroup_default, "mon_groups",
 					  &kn_mongrp);
 		if (ret < 0)
 			goto out_info;
@@ -2454,7 +2454,7 @@ static int mkdir_mondata_all(struct kernfs_node *parent_kn,
 	/*
 	 * Create the mon_data directory first.
 	 */
-	ret = mongroup_create_dir(parent_kn, NULL, "mon_data", &kn);
+	ret = mongroup_create_dir(parent_kn, prgrp, "mon_data", &kn);
 	if (ret)
 		return ret;
 
@@ -2653,7 +2653,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 	uint files = 0;
 	int ret;
 
-	prdtgrp = rdtgroup_kn_lock_live(prgrp_kn);
+	prdtgrp = rdtgroup_kn_lock_live(parent_kn);
 	if (!prdtgrp) {
 		ret = -ENODEV;
 		goto out_unlock;
@@ -2726,7 +2726,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 	kernfs_activate(kn);
 
 	/*
-	 * The caller unlocks the prgrp_kn upon success.
+	 * The caller unlocks the parent_kn upon success.
 	 */
 	return 0;
 
@@ -2737,7 +2737,7 @@ out_destroy:
 out_free_rgrp:
 	kfree(rdtgrp);
 out_unlock:
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
@@ -2775,7 +2775,7 @@ static int rdtgroup_mkdir_mon(struct kernfs_node *parent_kn,
 	 */
 	list_add_tail(&rdtgrp->mon.crdtgrp_list, &prgrp->mon.crdtgrp_list);
 
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
@@ -2818,7 +2818,7 @@ static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
 		 * Create an empty mon_groups directory to hold the subset
 		 * of tasks and cpus to monitor.
 		 */
-		ret = mongroup_create_dir(kn, NULL, "mon_groups", NULL);
+		ret = mongroup_create_dir(kn, rdtgrp, "mon_groups", NULL);
 		if (ret) {
 			rdt_last_cmd_puts("kernfs subdir error\n");
 			goto out_del_list;
@@ -2834,7 +2834,7 @@ out_id_free:
 out_common_fail:
 	mkdir_rdt_prepare_clean(rdtgrp);
 out_unlock:
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 

A message that is not in the archive (sent off-list or not yet received): [email protected]

maintainer
#14 lore raw

Re: [tip: x86/urgent] x86/resctrl: Fix a deadlock due to inaccurate reference

+Xiaochen (author of the patch)

Quoted 21 lines · Sasha Levin

Hi,

[This is an automated email]

This commit has been processed because it contains a "Fixes:" tag,
fixing commit: c7d9aac61311 ("x86/intel_rdt/cqm: Add mkdir support for RDT monitoring").

The bot has tested the following trees: v5.4.13, v4.19.97, v4.14.166.

v5.4.13: Build OK!
v4.19.97: Failed to apply! Possible dependencies:
Unable to calculate

v4.14.166: Failed to apply! Possible dependencies:
cfd0f34e4cd5 ("x86/intel_rdt: Add diagnostics when making directories")

NOTE: The patch will not be queued to stable trees until it is upstream.

How should we proceed with this patch?

#15 lore raw

PATCH 4/4 x86/resctrl: Clean up unused function parameter in mkdir path

In reply to Xiaochen Shen

Previous patch ("x86/resctrl: Fix a deadlock due to inaccurate active reference of kernfs node") changed the argument to rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() within mkdir_rdt_prepare(). That change resulted in an unused function parameter to mkdir_rdt_prepare().

Clean up unused function parameter in mkdir_rdt_prepare() and its callers rdtgroup_mkdir_mon() and rdtgroup_mkdir_ctrl_mon().

---
 arch/x86/kernel/cpu/resctrl/rdtgroup.c | 16 +++++-----------
 1 file changed, 5 insertions(+), 11 deletions(-)
arch/x86/kernel/cpu/resctrl/rdtgroup.c +5 -11
index 2c79fcd83271..8ebef16e10be 100644
@@ -2648,7 +2648,6 @@ static int rdtgroup_init_alloc(struct rdtgroup *rdtgrp)
 }
 
 static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
-			     struct kernfs_node *prgrp_kn,
 			     const char *name, umode_t mode,
 			     enum rdt_group_type rtype, struct rdtgroup **r)
 {
@@ -2758,15 +2757,12 @@ static void mkdir_rdt_prepare_clean(struct rdtgroup *rgrp)
  * to monitor a subset of tasks and cpus in its parent ctrl_mon group.
  */
 static int rdtgroup_mkdir_mon(struct kernfs_node *parent_kn,
-			      struct kernfs_node *prgrp_kn,
-			      const char *name,
-			      umode_t mode)
+			      const char *name, umode_t mode)
 {
 	struct rdtgroup *rdtgrp, *prgrp;
 	int ret;
 
-	ret = mkdir_rdt_prepare(parent_kn, prgrp_kn, name, mode, RDTMON_GROUP,
-				&rdtgrp);
+	ret = mkdir_rdt_prepare(parent_kn, name, mode, RDTMON_GROUP, &rdtgrp);
 	if (ret)
 		return ret;
 
@@ -2788,7 +2784,6 @@ static int rdtgroup_mkdir_mon(struct kernfs_node *parent_kn,
  * to allocate and monitor resources.
  */
 static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
-				   struct kernfs_node *prgrp_kn,
 				   const char *name, umode_t mode)
 {
 	struct rdtgroup *rdtgrp;
@@ -2796,8 +2791,7 @@ static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
 	u32 closid;
 	int ret;
 
-	ret = mkdir_rdt_prepare(parent_kn, prgrp_kn, name, mode, RDTCTRL_GROUP,
-				&rdtgrp);
+	ret = mkdir_rdt_prepare(parent_kn, name, mode, RDTCTRL_GROUP, &rdtgrp);
 	if (ret)
 		return ret;
 
@@ -2871,14 +2865,14 @@ static int rdtgroup_mkdir(struct kernfs_node *parent_kn, const char *name,
 	 * subdirectory
 	 */
 	if (rdt_alloc_capable && parent_kn == rdtgroup_default.kn)
-		return rdtgroup_mkdir_ctrl_mon(parent_kn, parent_kn, name, mode);
+		return rdtgroup_mkdir_ctrl_mon(parent_kn, name, mode);
 
 	/*
 	 * If RDT monitoring is supported and the parent directory is a valid
 	 * "mon_groups" directory, add a monitoring subdirectory.
 	 */
 	if (rdt_mon_capable && is_mon_groups(parent_kn, name))
-		return rdtgroup_mkdir_mon(parent_kn, parent_kn->parent, name, mode);
+		return rdtgroup_mkdir_mon(parent_kn, name, mode);
 
 	return -EPERM;
 }
Signature · 1 line
1.8.3.1
merge bot Applied
#16 lore raw

[tip: x86/urgent] x86/resctrl: Clean up unused function parameter in mkdir path

The following commit has been merged into the x86/urgent branch of tip:

The following commit has been merged into the x86/urgent branch of tip:

Commit-ID:     32ada3b9e04c6f6d4916967bd8bbe2450ad5bc5e
Gitweb:        https://git.kernel.org/tip/32ada3b9e04c6f6d4916967bd8bbe2450ad5bc5e
Author:        Xiaochen Shen <[email protected]>
AuthorDate:    Thu, 09 Jan 2020 00:28:06 +08:00
Committer:     Borislav Petkov <[email protected]>
CommitterDate: Mon, 20 Jan 2020 17:00:41 +01:00

x86/resctrl: Clean up unused function parameter in mkdir path

Commit

334b0f4e9b1b ("x86/resctrl: Fix a deadlock due to inaccurate reference")

changed the argument to rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() within mkdir_rdt_prepare(). That change resulted in an unused function parameter to mkdir_rdt_prepare().

Clean up the unused function parameter in mkdir_rdt_prepare() and its callers rdtgroup_mkdir_mon() and rdtgroup_mkdir_ctrl_mon().

---
 arch/x86/kernel/cpu/resctrl/rdtgroup.c | 16 +++++-----------
 1 file changed, 5 insertions(+), 11 deletions(-)
arch/x86/kernel/cpu/resctrl/rdtgroup.c +5 -11
index 954fd04..2804562 100644
@@ -2644,7 +2644,6 @@ static int rdtgroup_init_alloc(struct rdtgroup *rdtgrp)
 }
 
 static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
-			     struct kernfs_node *prgrp_kn,
 			     const char *name, umode_t mode,
 			     enum rdt_group_type rtype, struct rdtgroup **r)
 {
@@ -2754,15 +2753,12 @@ static void mkdir_rdt_prepare_clean(struct rdtgroup *rgrp)
  * to monitor a subset of tasks and cpus in its parent ctrl_mon group.
  */
 static int rdtgroup_mkdir_mon(struct kernfs_node *parent_kn,
-			      struct kernfs_node *prgrp_kn,
-			      const char *name,
-			      umode_t mode)
+			      const char *name, umode_t mode)
 {
 	struct rdtgroup *rdtgrp, *prgrp;
 	int ret;
 
-	ret = mkdir_rdt_prepare(parent_kn, prgrp_kn, name, mode, RDTMON_GROUP,
-				&rdtgrp);
+	ret = mkdir_rdt_prepare(parent_kn, name, mode, RDTMON_GROUP, &rdtgrp);
 	if (ret)
 		return ret;
 
@@ -2784,7 +2780,6 @@ static int rdtgroup_mkdir_mon(struct kernfs_node *parent_kn,
  * to allocate and monitor resources.
  */
 static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
-				   struct kernfs_node *prgrp_kn,
 				   const char *name, umode_t mode)
 {
 	struct rdtgroup *rdtgrp;
@@ -2792,8 +2787,7 @@ static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
 	u32 closid;
 	int ret;
 
-	ret = mkdir_rdt_prepare(parent_kn, prgrp_kn, name, mode, RDTCTRL_GROUP,
-				&rdtgrp);
+	ret = mkdir_rdt_prepare(parent_kn, name, mode, RDTCTRL_GROUP, &rdtgrp);
 	if (ret)
 		return ret;
 
@@ -2867,14 +2861,14 @@ static int rdtgroup_mkdir(struct kernfs_node *parent_kn, const char *name,
 	 * subdirectory
 	 */
 	if (rdt_alloc_capable && parent_kn == rdtgroup_default.kn)
-		return rdtgroup_mkdir_ctrl_mon(parent_kn, parent_kn, name, mode);
+		return rdtgroup_mkdir_ctrl_mon(parent_kn, name, mode);
 
 	/*
 	 * If RDT monitoring is supported and the parent directory is a valid
 	 * "mon_groups" directory, add a monitoring subdirectory.
 	 */
 	if (rdt_mon_capable && is_mon_groups(parent_kn, name))
-		return rdtgroup_mkdir_mon(parent_kn, parent_kn->parent, name, mode);
+		return rdtgroup_mkdir_mon(parent_kn, name, mode);
 
 	return -EPERM;
 }
#17 lore raw

PATCH 1/3 x86/resctrl: Fix use-after-free when deleting resource groups

In reply to Xiaochen Shen

commit b8511ccc75c033f6d54188ea4df7bf1e85778740 upstream.

A resource group (rdtgrp) contains a reference count (rdtgrp->waitcount) that indicates how many waiters expect this rdtgrp to exist. Waiters could be waiting on rdtgroup_mutex or some work sitting on a task's workqueue for when the task returns from kernel mode or exits.

The deletion of a rdtgrp is intended to have two phases:

(1) while holding rdtgroup_mutex the necessary cleanup is done and rdtgrp->flags is set to RDT_DELETED,

(2) after releasing the rdtgroup_mutex, the rdtgrp structure is freed only if there are no waiters and its flag is set to RDT_DELETED. Upon gaining access to rdtgroup_mutex or rdtgrp, a waiter is required to check for the RDT_DELETED flag.

When unmounting the resctrl file system or deleting ctrl_mon groups, all of the subdirectories are removed and the data structure of rdtgrp is forcibly freed without checking rdtgrp->waitcount. If at this point there was a waiter on rdtgrp then a use-after-free issue occurs when the waiter starts running and accesses the rdtgrp structure it was waiting on.

See kfree() calls in [1], [2] and [3] in these two call paths in following scenarios:
(1) rdt_kill_sb() -> rmdir_all_sub() -> free_all_child_rdtgrp()
(2) rdtgroup_rmdir() -> rdtgroup_rmdir_ctrl() -> free_all_child_rdtgrp()

There are several scenarios that result in use-after-free issue in following:

Scenario 1:
-----------
In Thread 1, rdtgroup_tasks_write() adds a task_work callback move_myself(). If move_myself() is scheduled to execute after Thread 2 rdt_kill_sb() is finished, referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_tasks_write) Thread 2 (rdt_kill_sb)
------------------------------- ----------------------
rdtgroup_kn_lock_live
atomic_inc(&rdtgrp->waitcount)
mutex_lock
rdtgroup_move_task
__rdtgroup_move_task

    /*
     * Take an extra refcount, so rdtgrp cannot be freed
     * before the call back move_myself has been invoked
     */
    atomic_inc(&rdtgrp->waitcount)
    /* Callback move_myself will be scheduled for later */
    task_work_add(move_myself)

rdtgroup_kn_unlock
mutex_unlock
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

                                       mutex_lock
                                       rmdir_all_sub
                                         /*
                                          * sentry and rdtgrp are freed
                                          * without checking refcount
                                          */
                                         free_all_child_rdtgrp
                                           kfree(sentry)*[1]
                                         kfree(rdtgrp)*[2]
                                       mutex_unlock

/*
* Callback is scheduled to execute
* after rdt_kill_sb is finished
*/
move_myself
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [1] or [2].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)
kfree(rdtgrp)

Scenario 2:
-----------
In Thread 1, rdtgroup_tasks_write() adds a task_work callback move_myself(). If move_myself() is scheduled to execute after Thread 2 rdtgroup_rmdir() is finished, referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_tasks_write) Thread 2 (rdtgroup_rmdir)
------------------------------- -------------------------
rdtgroup_kn_lock_live
atomic_inc(&rdtgrp->waitcount)
mutex_lock
rdtgroup_move_task
__rdtgroup_move_task

    /*
     * Take an extra refcount, so rdtgrp cannot be freed
     * before the call back move_myself has been invoked
     */
    atomic_inc(&rdtgrp->waitcount)
    /* Callback move_myself will be scheduled for later */
    task_work_add(move_myself)

rdtgroup_kn_unlock
mutex_unlock
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

                                       rdtgroup_kn_lock_live
                                         atomic_inc(&rdtgrp->waitcount)
                                         mutex_lock
                                       rdtgroup_rmdir_ctrl
                                         free_all_child_rdtgrp
                                           /*
                                            * sentry is freed without
                                            * checking refcount
                                            */
                                           kfree(sentry)*[3]
                                         rdtgroup_ctrl_remove
                                           rdtgrp->flags = RDT_DELETED
                                       rdtgroup_kn_unlock
                                         mutex_unlock
                                         atomic_dec_and_test(
                                                     &rdtgrp->waitcount)
                                         && (flags & RDT_DELETED)
                                           kfree(rdtgrp)

/*
* Callback is scheduled to execute
* after rdt_kill_sb is finished
*/
move_myself
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [3].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)
kfree(rdtgrp)

If CONFIG_DEBUG_SLAB=y, Slab corruption on kmalloc-2k can be observed like following. Note that "0x6b" is POISON_FREE after kfree(). The corrupted bits "0x6a", "0x64" at offset 0x424 correspond to waitcount member of struct rdtgroup which was freed:

Slab corruption (Not tainted): kmalloc-2k start=ffff9504c5b0d000, len=2048 420: 6b 6b 6b 6b 6a 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkjkkkkkkkkkkk Single bit error detected. Probably bad RAM.
Run memtest86+ or a similar memory test tool.
Next obj: start=ffff9504c5b0d800, len=2048
000: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk 010: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk

Slab corruption (Not tainted): kmalloc-2k start=ffff9504c58ab800, len=2048 420: 6b 6b 6b 6b 64 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkdkkkkkkkkkkk Prev obj: start=ffff9504c58ab000, len=2048
000: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk 010: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk

Fix this by taking reference count (waitcount) of rdtgrp into account in the two call paths that currently do not do so. Instead of always freeing the resource group it will only be freed if there are no waiters on it. If there are waiters, the resource group will have its flags set to RDT_DELETED.

It will be left to the waiter to free the resource group when it starts running and finding that it was the last waiter and the resource group has been removed (rdtgrp->flags & RDT_DELETED) since. (1) rdt_kill_sb() -> rmdir_all_sub() -> free_all_child_rdtgrp() (2) rdtgroup_rmdir() -> rdtgroup_rmdir_ctrl() -> free_all_child_rdtgrp()

Backporting notes:

Since upstream commit fa7d949337cc ("x86/resctrl: Rename and move rdt files to a separate directory"), the file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c has been renamed and moved to arch/x86/kernel/cpu/resctrl/rdtgroup.c.

Apply the change against file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c in older stable trees.

Fixes f3cbeacaa06e ("x86/intel_rdt/cqm: Add rmdir support")
Fixes 60cf5e101fd4 ("x86/intel_rdt: Add mkdir to resctrl file system")
---
 arch/x86/kernel/cpu/intel_rdt_rdtgroup.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
arch/x86/kernel/cpu/intel_rdt_rdtgroup.c +10 -2
index 2dae1b3..7349969 100644
@@ -1260,7 +1260,11 @@ static void free_all_child_rdtgrp(struct rdtgroup *rdtgrp)
 	list_for_each_entry_safe(sentry, stmp, head, mon.crdtgrp_list) {
 		free_rmid(sentry->mon.rmid);
 		list_del(&sentry->mon.crdtgrp_list);
-		kfree(sentry);
+
+		if (atomic_read(&sentry->waitcount) != 0)
+			sentry->flags = RDT_DELETED;
+		else
+			kfree(sentry);
 	}
 }
 
@@ -1294,7 +1298,11 @@ static void rmdir_all_sub(void)
 
 		kernfs_remove(rdtgrp->kn);
 		list_del(&rdtgrp->rdtgroup_list);
-		kfree(rdtgrp);
+
+		if (atomic_read(&rdtgrp->waitcount) != 0)
+			rdtgrp->flags = RDT_DELETED;
+		else
+			kfree(rdtgrp);
 	}
 	/* Notify online CPUs to update per cpu storage and PQR_ASSOC MSR */
 	update_closid_rmid(cpu_online_mask, &rdtgroup_default);
Signature · 1 line
1.8.3.1
#18 lore raw

PATCH 1/3 x86/resctrl: Fix use-after-free when deleting resource groups

In reply to Xiaochen Shen

commit b8511ccc75c033f6d54188ea4df7bf1e85778740 upstream.

A resource group (rdtgrp) contains a reference count (rdtgrp->waitcount) that indicates how many waiters expect this rdtgrp to exist. Waiters could be waiting on rdtgroup_mutex or some work sitting on a task's workqueue for when the task returns from kernel mode or exits.

The deletion of a rdtgrp is intended to have two phases:

(1) while holding rdtgroup_mutex the necessary cleanup is done and rdtgrp->flags is set to RDT_DELETED,

(2) after releasing the rdtgroup_mutex, the rdtgrp structure is freed only if there are no waiters and its flag is set to RDT_DELETED. Upon gaining access to rdtgroup_mutex or rdtgrp, a waiter is required to check for the RDT_DELETED flag.

When unmounting the resctrl file system or deleting ctrl_mon groups, all of the subdirectories are removed and the data structure of rdtgrp is forcibly freed without checking rdtgrp->waitcount. If at this point there was a waiter on rdtgrp then a use-after-free issue occurs when the waiter starts running and accesses the rdtgrp structure it was waiting on.

See kfree() calls in [1], [2] and [3] in these two call paths in following scenarios:
(1) rdt_kill_sb() -> rmdir_all_sub() -> free_all_child_rdtgrp()
(2) rdtgroup_rmdir() -> rdtgroup_rmdir_ctrl() -> free_all_child_rdtgrp()

There are several scenarios that result in use-after-free issue in following:

Scenario 1:
-----------
In Thread 1, rdtgroup_tasks_write() adds a task_work callback move_myself(). If move_myself() is scheduled to execute after Thread 2 rdt_kill_sb() is finished, referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_tasks_write) Thread 2 (rdt_kill_sb)
------------------------------- ----------------------
rdtgroup_kn_lock_live
atomic_inc(&rdtgrp->waitcount)
mutex_lock
rdtgroup_move_task
__rdtgroup_move_task

    /*
     * Take an extra refcount, so rdtgrp cannot be freed
     * before the call back move_myself has been invoked
     */
    atomic_inc(&rdtgrp->waitcount)
    /* Callback move_myself will be scheduled for later */
    task_work_add(move_myself)

rdtgroup_kn_unlock
mutex_unlock
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

                                       mutex_lock
                                       rmdir_all_sub
                                         /*
                                          * sentry and rdtgrp are freed
                                          * without checking refcount
                                          */
                                         free_all_child_rdtgrp
                                           kfree(sentry)*[1]
                                         kfree(rdtgrp)*[2]
                                       mutex_unlock

/*
* Callback is scheduled to execute
* after rdt_kill_sb is finished
*/
move_myself
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [1] or [2].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)
kfree(rdtgrp)

Scenario 2:
-----------
In Thread 1, rdtgroup_tasks_write() adds a task_work callback move_myself(). If move_myself() is scheduled to execute after Thread 2 rdtgroup_rmdir() is finished, referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_tasks_write) Thread 2 (rdtgroup_rmdir)
------------------------------- -------------------------
rdtgroup_kn_lock_live
atomic_inc(&rdtgrp->waitcount)
mutex_lock
rdtgroup_move_task
__rdtgroup_move_task

    /*
     * Take an extra refcount, so rdtgrp cannot be freed
     * before the call back move_myself has been invoked
     */
    atomic_inc(&rdtgrp->waitcount)
    /* Callback move_myself will be scheduled for later */
    task_work_add(move_myself)

rdtgroup_kn_unlock
mutex_unlock
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

                                       rdtgroup_kn_lock_live
                                         atomic_inc(&rdtgrp->waitcount)
                                         mutex_lock
                                       rdtgroup_rmdir_ctrl
                                         free_all_child_rdtgrp
                                           /*
                                            * sentry is freed without
                                            * checking refcount
                                            */
                                           kfree(sentry)*[3]
                                         rdtgroup_ctrl_remove
                                           rdtgrp->flags = RDT_DELETED
                                       rdtgroup_kn_unlock
                                         mutex_unlock
                                         atomic_dec_and_test(
                                                     &rdtgrp->waitcount)
                                         && (flags & RDT_DELETED)
                                           kfree(rdtgrp)

/*
* Callback is scheduled to execute
* after rdt_kill_sb is finished
*/
move_myself
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [3].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)
kfree(rdtgrp)

If CONFIG_DEBUG_SLAB=y, Slab corruption on kmalloc-2k can be observed like following. Note that "0x6b" is POISON_FREE after kfree(). The corrupted bits "0x6a", "0x64" at offset 0x424 correspond to waitcount member of struct rdtgroup which was freed:

Slab corruption (Not tainted): kmalloc-2k start=ffff9504c5b0d000, len=2048 420: 6b 6b 6b 6b 6a 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkjkkkkkkkkkkk Single bit error detected. Probably bad RAM.
Run memtest86+ or a similar memory test tool.
Next obj: start=ffff9504c5b0d800, len=2048
000: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk 010: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk

Slab corruption (Not tainted): kmalloc-2k start=ffff9504c58ab800, len=2048 420: 6b 6b 6b 6b 64 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkdkkkkkkkkkkk Prev obj: start=ffff9504c58ab000, len=2048
000: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk 010: 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b kkkkkkkkkkkkkkkk

Fix this by taking reference count (waitcount) of rdtgrp into account in the two call paths that currently do not do so. Instead of always freeing the resource group it will only be freed if there are no waiters on it. If there are waiters, the resource group will have its flags set to RDT_DELETED.

It will be left to the waiter to free the resource group when it starts running and finding that it was the last waiter and the resource group has been removed (rdtgrp->flags & RDT_DELETED) since. (1) rdt_kill_sb() -> rmdir_all_sub() -> free_all_child_rdtgrp() (2) rdtgroup_rmdir() -> rdtgroup_rmdir_ctrl() -> free_all_child_rdtgrp()

Backporting notes:

Since upstream commit fa7d949337cc ("x86/resctrl: Rename and move rdt files to a separate directory"), the file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c has been renamed and moved to arch/x86/kernel/cpu/resctrl/rdtgroup.c.

Apply the change against file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c in older stable trees.

Fixes f3cbeacaa06e ("x86/intel_rdt/cqm: Add rmdir support")
Fixes 60cf5e101fd4 ("x86/intel_rdt: Add mkdir to resctrl file system")
---
 arch/x86/kernel/cpu/intel_rdt_rdtgroup.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
arch/x86/kernel/cpu/intel_rdt_rdtgroup.c +10 -2
index 841a024..db22ba0 100644
@@ -2167,7 +2167,11 @@ static void free_all_child_rdtgrp(struct rdtgroup *rdtgrp)
 	list_for_each_entry_safe(sentry, stmp, head, mon.crdtgrp_list) {
 		free_rmid(sentry->mon.rmid);
 		list_del(&sentry->mon.crdtgrp_list);
-		kfree(sentry);
+
+		if (atomic_read(&sentry->waitcount) != 0)
+			sentry->flags = RDT_DELETED;
+		else
+			kfree(sentry);
 	}
 }
 
@@ -2205,7 +2209,11 @@ static void rmdir_all_sub(void)
 
 		kernfs_remove(rdtgrp->kn);
 		list_del(&rdtgrp->rdtgroup_list);
-		kfree(rdtgrp);
+
+		if (atomic_read(&rdtgrp->waitcount) != 0)
+			rdtgrp->flags = RDT_DELETED;
+		else
+			kfree(rdtgrp);
 	}
 	/* Notify online CPUs to update per cpu storage and PQR_ASSOC MSR */
 	update_closid_rmid(cpu_online_mask, &rdtgroup_default);
Signature · 1 line
1.8.3.1
maintainer Applied
#19 lore raw

Re: [PATCH 4.19 1/3] x86/resctrl: Fix use-after-free when deleting resource groups

Quoted 6 lines · Xiaochen Shen

commit b8511ccc75c033f6d54188ea4df7bf1e85778740 upstream.

A resource group (rdtgrp) contains a reference count (rdtgrp->waitcount) that indicates how many waiters expect this rdtgrp to exist. Waiters could be waiting on rdtgroup_mutex or some work sitting on a task's workqueue for when the task returns from kernel mode or exits.

I've queued up the 4.19 and 4.14 backports, thanks!

Signature · 2 lines
Thanks,
Sasha
#20 lore raw

PATCH 2/3 x86/resctrl: Fix use-after-free due to inaccurate refcount of rdtgroup

In reply to Xiaochen Shen

commit 074fadee59ee7a9d2b216e9854bd4efb5dad679f upstream.

There is a race condition in the following scenario which results in an use-after-free issue when reading a monitoring file and deleting the parent ctrl_mon group concurrently:

Thread 1 calls atomic_inc() to take refcount of rdtgrp and then calls kernfs_break_active_protection() to drop the active reference of kernfs node in rdtgroup_kn_lock_live().

In Thread 2, kernfs_remove() is a blocking routine. It waits on all sub kernfs nodes to drop the active reference when removing all subtree kernfs nodes recursively. Thread 2 could block on kernfs_remove() until Thread 1 calls kernfs_break_active_protection(). Only after kernfs_remove() completes the refcount of rdtgrp could be trusted.

Before Thread 1 calls atomic_inc() and kernfs_break_active_protection(), Thread 2 could call kfree() when the refcount of rdtgrp (sentry) is 0 instead of 1 due to the race.

In Thread 1, in rdtgroup_kn_unlock(), referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_mondata_show) Thread 2 (rdtgroup_rmdir) -------------------------------- ------------------------- rdtgroup_kn_lock_live
/*
* kn active protection until
* kernfs_break_active_protection(kn)
*/
rdtgrp = kernfs_to_rdtgroup(kn)

                                  rdtgroup_kn_lock_live
                                    atomic_inc(&rdtgrp->waitcount)
                                    mutex_lock
                                  rdtgroup_rmdir_ctrl
                                    free_all_child_rdtgrp
                                      /*
                                       * sentry->waitcount should be 1
                                       * but is 0 now due to the race.
                                       */
                                      kfree(sentry)*[1]
  /*
   * Only after kernfs_remove()
   * completes, the refcount of
   * rdtgrp could be trusted.
   */
  atomic_inc(&rdtgrp->waitcount)
  /* kn->active-- */
  kernfs_break_active_protection(kn)
                                    rdtgroup_ctrl_remove
                                      rdtgrp->flags = RDT_DELETED
                                      /*
                                       * Blocking routine, wait for
                                       * all sub kernfs nodes to drop
                                       * active reference in
                                       * kernfs_break_active_protection.
                                       */
                                      kernfs_remove(rdtgrp->kn)
                                  rdtgroup_kn_unlock
                                    mutex_unlock
                                    atomic_dec_and_test(
                                                &rdtgrp->waitcount)
                                    && (flags & RDT_DELETED)
                                      kernfs_unbreak_active_protection(kn)
                                      kfree(rdtgrp)

mutex_lock
mon_event_read
rdtgroup_kn_unlock
mutex_unlock
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [1].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

    /* kn->active++ */
    kernfs_unbreak_active_protection(kn)
    kfree(rdtgrp)

Fix it by moving free_all_child_rdtgrp() to after kernfs_remove() in rdtgroup_rmdir_ctrl() to ensure it has the accurate refcount of rdtgrp.

Backporting notes:

Since upstream commit fa7d949337cc ("x86/resctrl: Rename and move rdt files to a separate directory"), the file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c has been renamed and moved to arch/x86/kernel/cpu/resctrl/rdtgroup.c.
Apply the change against file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c for older stable trees.

Upstream commit 17eafd076291 ("x86/intel_rdt: Split resource group removal in two") moved part of resource group removal code from rdtgroup_rmdir_mon() into a separate function rdtgroup_ctrl_remove(). Apply the change against original code base of rdtgroup_rmdir_mon() for older stable trees.

---
 arch/x86/kernel/cpu/intel_rdt_rdtgroup.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
arch/x86/kernel/cpu/intel_rdt_rdtgroup.c +5 -5
index 7349969..0157496 100644
@@ -1800,11 +1800,6 @@ static int rdtgroup_rmdir_ctrl(struct kernfs_node *kn, struct rdtgroup *rdtgrp,
 	closid_free(rdtgrp->closid);
 	free_rmid(rdtgrp->mon.rmid);
 
-	/*
-	 * Free all the child monitor group rmids.
-	 */
-	free_all_child_rdtgrp(rdtgrp);
-
 	list_del(&rdtgrp->rdtgroup_list);
 
 	/*
@@ -1814,6 +1809,11 @@ static int rdtgroup_rmdir_ctrl(struct kernfs_node *kn, struct rdtgroup *rdtgrp,
 	kernfs_get(kn);
 	kernfs_remove(rdtgrp->kn);
 
+	/*
+	 * Free all the child monitor group rmids.
+	 */
+	free_all_child_rdtgrp(rdtgrp);
+
 	return 0;
 }
 
Signature · 1 line
1.8.3.1
#21 lore raw

PATCH 2/3 x86/resctrl: Fix use-after-free due to inaccurate refcount of rdtgroup

In reply to Xiaochen Shen

commit 074fadee59ee7a9d2b216e9854bd4efb5dad679f upstream.

There is a race condition in the following scenario which results in an use-after-free issue when reading a monitoring file and deleting the parent ctrl_mon group concurrently:

Thread 1 calls atomic_inc() to take refcount of rdtgrp and then calls kernfs_break_active_protection() to drop the active reference of kernfs node in rdtgroup_kn_lock_live().

In Thread 2, kernfs_remove() is a blocking routine. It waits on all sub kernfs nodes to drop the active reference when removing all subtree kernfs nodes recursively. Thread 2 could block on kernfs_remove() until Thread 1 calls kernfs_break_active_protection(). Only after kernfs_remove() completes the refcount of rdtgrp could be trusted.

Before Thread 1 calls atomic_inc() and kernfs_break_active_protection(), Thread 2 could call kfree() when the refcount of rdtgrp (sentry) is 0 instead of 1 due to the race.

In Thread 1, in rdtgroup_kn_unlock(), referring to earlier rdtgrp memory (rdtgrp->waitcount) which was already freed in Thread 2 results in use-after-free issue.

Thread 1 (rdtgroup_mondata_show) Thread 2 (rdtgroup_rmdir) -------------------------------- ------------------------- rdtgroup_kn_lock_live
/*
* kn active protection until
* kernfs_break_active_protection(kn)
*/
rdtgrp = kernfs_to_rdtgroup(kn)

                                  rdtgroup_kn_lock_live
                                    atomic_inc(&rdtgrp->waitcount)
                                    mutex_lock
                                  rdtgroup_rmdir_ctrl
                                    free_all_child_rdtgrp
                                      /*
                                       * sentry->waitcount should be 1
                                       * but is 0 now due to the race.
                                       */
                                      kfree(sentry)*[1]
  /*
   * Only after kernfs_remove()
   * completes, the refcount of
   * rdtgrp could be trusted.
   */
  atomic_inc(&rdtgrp->waitcount)
  /* kn->active-- */
  kernfs_break_active_protection(kn)
                                    rdtgroup_ctrl_remove
                                      rdtgrp->flags = RDT_DELETED
                                      /*
                                       * Blocking routine, wait for
                                       * all sub kernfs nodes to drop
                                       * active reference in
                                       * kernfs_break_active_protection.
                                       */
                                      kernfs_remove(rdtgrp->kn)
                                  rdtgroup_kn_unlock
                                    mutex_unlock
                                    atomic_dec_and_test(
                                                &rdtgrp->waitcount)
                                    && (flags & RDT_DELETED)
                                      kernfs_unbreak_active_protection(kn)
                                      kfree(rdtgrp)

mutex_lock
mon_event_read
rdtgroup_kn_unlock
mutex_unlock
/*
* Use-after-free: refer to earlier rdtgrp
* memory which was freed in [1].
*/
atomic_dec_and_test(&rdtgrp->waitcount)
&& (flags & RDT_DELETED)

    /* kn->active++ */
    kernfs_unbreak_active_protection(kn)
    kfree(rdtgrp)

Fix it by moving free_all_child_rdtgrp() to after kernfs_remove() in rdtgroup_rmdir_ctrl() to ensure it has the accurate refcount of rdtgrp.

Backporting notes:

Since upstream commit fa7d949337cc ("x86/resctrl: Rename and move rdt files to a separate directory"), the file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c has been renamed and moved to arch/x86/kernel/cpu/resctrl/rdtgroup.c.
Apply the change against file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c for older stable trees.

---
 arch/x86/kernel/cpu/intel_rdt_rdtgroup.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/x86/kernel/cpu/intel_rdt_rdtgroup.c +2 -2
index db22ba0..77770ca 100644
@@ -2877,13 +2877,13 @@ static int rdtgroup_rmdir_ctrl(struct kernfs_node *kn, struct rdtgroup *rdtgrp,
 	closid_free(rdtgrp->closid);
 	free_rmid(rdtgrp->mon.rmid);
 
+	rdtgroup_ctrl_remove(kn, rdtgrp);
+
 	/*
 	 * Free all the child monitor group rmids.
 	 */
 	free_all_child_rdtgrp(rdtgrp);
 
-	rdtgroup_ctrl_remove(kn, rdtgrp);
-
 	return 0;
 }
 
Signature · 1 line
1.8.3.1
#22 lore raw

PATCH 3/3 x86/resctrl: Fix a deadlock due to inaccurate reference

In reply to Xiaochen Shen

commit 334b0f4e9b1b4a1d475f803419d202f6c5e4d18e upstream.

There is a race condition which results in a deadlock when rmdir and mkdir execute concurrently:

$ ls /sys/fs/resctrl/c1/mon_groups/m1/
cpus  cpus_list  mon_data  tasks

Thread 1: rmdir /sys/fs/resctrl/c1
Thread 2: mkdir /sys/fs/resctrl/c1/mon_groups/m1

3 locks held by mkdir/48649:

 #0:  (sb_writers#17){.+.+}, at: [<ffffffffb4ca2aa0>] mnt_want_write+0x20/0x50
 #1:  (&type->i_mutex_dir_key#8/1){+.+.}, at: [<ffffffffb4c8c13b>] filename_create+0x7b/0x170
 #2:  (rdtgroup_mutex){+.+.}, at: [<ffffffffb4a4389d>] rdtgroup_kn_lock_live+0x3d/0x70

4 locks held by rmdir/48652:

 #0:  (sb_writers#17){.+.+}, at: [<ffffffffb4ca2aa0>] mnt_want_write+0x20/0x50
 #1:  (&type->i_mutex_dir_key#8/1){+.+.}, at: [<ffffffffb4c8c3cf>] do_rmdir+0x13f/0x1e0
 #2:  (&type->i_mutex_dir_key#8){++++}, at: [<ffffffffb4c86d5d>] vfs_rmdir+0x4d/0x120
 #3:  (rdtgroup_mutex){+.+.}, at: [<ffffffffb4a4389d>] rdtgroup_kn_lock_live+0x3d/0x70

Thread 1 is deleting control group "c1". Holding rdtgroup_mutex, kernfs_remove() removes all kernfs nodes under directory "c1" recursively, then waits for sub kernfs node "mon_groups" to drop active reference.

Thread 2 is trying to create a subdirectory "m1" in the "mon_groups" directory. The wrapper kernfs_iop_mkdir() takes an active reference to the "mon_groups" directory but the code drops the active reference to the parent directory "c1" instead.

As a result, Thread 1 is blocked on waiting for active reference to drop and never release rdtgroup_mutex, while Thread 2 is also blocked on trying to get rdtgroup_mutex.

Thread 1 (rdtgroup_rmdir)   Thread 2 (rdtgroup_mkdir)
(rmdir /sys/fs/resctrl/c1)  (mkdir /sys/fs/resctrl/c1/mon_groups/m1)
-------------------------   -------------------------

kernfs_iop_mkdir
/*
* kn: "m1", parent_kn: "mon_groups",
* prgrp_kn: parent_kn->parent: "c1", *
* "mon_groups", parent_kn->active++: 1 */
kernfs_get_active(parent_kn)
kernfs_iop_rmdir
/* "c1", kn->active++ */
kernfs_get_active(kn)

rdtgroup_kn_lock_live

    atomic_inc(&rdtgrp->waitcount)
    /* "c1", kn->active-- */
    kernfs_break_active_protection(kn)
    mutex_lock

rdtgroup_rmdir_ctrl

    free_all_child_rdtgrp
      sentry->flags = RDT_DELETED

    rdtgroup_ctrl_remove
      rdtgrp->flags = RDT_DELETED
      kernfs_get(kn)
      kernfs_remove(rdtgrp->kn)
        __kernfs_remove
          /* "mon_groups", sub_kn */
          atomic_add(KN_DEACTIVATED_BIAS, &sub_kn->active)
          kernfs_drain(sub_kn)
            /*
             * sub_kn->active == KN_DEACTIVATED_BIAS + 1,
             * waiting on sub_kn->active to drop, but it
             * never drops in Thread 2 which is blocked
             * on getting rdtgroup_mutex.
             */

Thread 1 hangs here ---->
wait_event(sub_kn->active == KN_DEACTIVATED_BIAS)
...
rdtgroup_mkdir
rdtgroup_mkdir_mon(parent_kn, prgrp_kn)
mkdir_rdt_prepare(parent_kn, prgrp_kn)
rdtgroup_kn_lock_live(prgrp_kn)
atomic_inc(&rdtgrp->waitcount) /*
* "c1", prgrp_kn->active-- *
* The active reference on "c1" is
* dropped, but not matching the
* actual active reference taken
* on "mon_groups", thus causing
* Thread 1 to wait forever while
* holding rdtgroup_mutex. */
kernfs_break_active_protection(
prgrp_kn)
/*
* Trying to get rdtgroup_mutex
* which is held by Thread 1. */
Thread 2 hangs here ----> mutex_lock

                                      ...

The problem is that the creation of a subdirectory in the "mon_groups" directory incorrectly releases the active protection of its parent directory instead of itself before it starts waiting for rdtgroup_mutex. This is triggered by the rdtgroup_mkdir() flow calling rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() with kernfs node of the parent control group ("c1") as argument. It should be called with kernfs node "mon_groups" instead. What is currently missing is that the kn->priv of "mon_groups" is NULL instead of pointing to the rdtgrp.

Fix it by pointing kn->priv to rdtgrp when "mon_groups" is created. Then it could be passed to rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() instead. And then it operates on the same rdtgroup structure but handles the active reference of kernfs node "mon_groups" to prevent deadlock. The same changes are also made to the "mon_data" directories.

This results in some unused function parameters that will be cleaned up in follow-up patch as the focus here is on the fix only in support of backporting efforts.

Backporting notes:

Since upstream commit fa7d949337cc ("x86/resctrl: Rename and move rdt files to a separate directory"), the file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c has been renamed and moved to arch/x86/kernel/cpu/resctrl/rdtgroup.c.
Apply the change against file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c for older stable trees.

---
 arch/x86/kernel/cpu/intel_rdt_rdtgroup.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
arch/x86/kernel/cpu/intel_rdt_rdtgroup.c +8 -8
index 0157496..0ec30b2 100644
@@ -1107,7 +1107,7 @@ static struct dentry *rdt_mount(struct file_system_type *fs_type,
 
 	if (rdt_mon_capable) {
 		ret = mongroup_create_dir(rdtgroup_default.kn,
-					  NULL, "mon_groups",
+					  &rdtgroup_default, "mon_groups",
 					  &kn_mongrp);
 		if (ret) {
 			dentry = ERR_PTR(ret);
@@ -1499,7 +1499,7 @@ static int mkdir_mondata_all(struct kernfs_node *parent_kn,
 	/*
 	 * Create the mon_data directory first.
 	 */
-	ret = mongroup_create_dir(parent_kn, NULL, "mon_data", &kn);
+	ret = mongroup_create_dir(parent_kn, prgrp, "mon_data", &kn);
 	if (ret)
 		return ret;
 
@@ -1533,7 +1533,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 	uint files = 0;
 	int ret;
 
-	prdtgrp = rdtgroup_kn_lock_live(prgrp_kn);
+	prdtgrp = rdtgroup_kn_lock_live(parent_kn);
 	if (!prdtgrp) {
 		ret = -ENODEV;
 		goto out_unlock;
@@ -1589,7 +1589,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 	kernfs_activate(kn);
 
 	/*
-	 * The caller unlocks the prgrp_kn upon success.
+	 * The caller unlocks the parent_kn upon success.
 	 */
 	return 0;
 
@@ -1600,7 +1600,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 out_free_rgrp:
 	kfree(rdtgrp);
 out_unlock:
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
@@ -1638,7 +1638,7 @@ static int rdtgroup_mkdir_mon(struct kernfs_node *parent_kn,
 	 */
 	list_add_tail(&rdtgrp->mon.crdtgrp_list, &prgrp->mon.crdtgrp_list);
 
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
@@ -1675,7 +1675,7 @@ static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
 		 * Create an empty mon_groups directory to hold the subset
 		 * of tasks and cpus to monitor.
 		 */
-		ret = mongroup_create_dir(kn, NULL, "mon_groups", NULL);
+		ret = mongroup_create_dir(kn, rdtgrp, "mon_groups", NULL);
 		if (ret)
 			goto out_id_free;
 	}
@@ -1688,7 +1688,7 @@ static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
 out_common_fail:
 	mkdir_rdt_prepare_clean(rdtgrp);
 out_unlock:
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
Signature · 1 line
1.8.3.1
#23 lore raw

PATCH 3/3 x86/resctrl: Fix a deadlock due to inaccurate reference

In reply to Xiaochen Shen

commit 334b0f4e9b1b4a1d475f803419d202f6c5e4d18e upstream.

There is a race condition which results in a deadlock when rmdir and mkdir execute concurrently:

$ ls /sys/fs/resctrl/c1/mon_groups/m1/
cpus  cpus_list  mon_data  tasks

Thread 1: rmdir /sys/fs/resctrl/c1
Thread 2: mkdir /sys/fs/resctrl/c1/mon_groups/m1

3 locks held by mkdir/48649:

 #0:  (sb_writers#17){.+.+}, at: [<ffffffffb4ca2aa0>] mnt_want_write+0x20/0x50
 #1:  (&type->i_mutex_dir_key#8/1){+.+.}, at: [<ffffffffb4c8c13b>] filename_create+0x7b/0x170
 #2:  (rdtgroup_mutex){+.+.}, at: [<ffffffffb4a4389d>] rdtgroup_kn_lock_live+0x3d/0x70

4 locks held by rmdir/48652:

 #0:  (sb_writers#17){.+.+}, at: [<ffffffffb4ca2aa0>] mnt_want_write+0x20/0x50
 #1:  (&type->i_mutex_dir_key#8/1){+.+.}, at: [<ffffffffb4c8c3cf>] do_rmdir+0x13f/0x1e0
 #2:  (&type->i_mutex_dir_key#8){++++}, at: [<ffffffffb4c86d5d>] vfs_rmdir+0x4d/0x120
 #3:  (rdtgroup_mutex){+.+.}, at: [<ffffffffb4a4389d>] rdtgroup_kn_lock_live+0x3d/0x70

Thread 1 is deleting control group "c1". Holding rdtgroup_mutex, kernfs_remove() removes all kernfs nodes under directory "c1" recursively, then waits for sub kernfs node "mon_groups" to drop active reference.

Thread 2 is trying to create a subdirectory "m1" in the "mon_groups" directory. The wrapper kernfs_iop_mkdir() takes an active reference to the "mon_groups" directory but the code drops the active reference to the parent directory "c1" instead.

As a result, Thread 1 is blocked on waiting for active reference to drop and never release rdtgroup_mutex, while Thread 2 is also blocked on trying to get rdtgroup_mutex.

Thread 1 (rdtgroup_rmdir)   Thread 2 (rdtgroup_mkdir)
(rmdir /sys/fs/resctrl/c1)  (mkdir /sys/fs/resctrl/c1/mon_groups/m1)
-------------------------   -------------------------

kernfs_iop_mkdir
/*
* kn: "m1", parent_kn: "mon_groups",
* prgrp_kn: parent_kn->parent: "c1", *
* "mon_groups", parent_kn->active++: 1 */
kernfs_get_active(parent_kn)
kernfs_iop_rmdir
/* "c1", kn->active++ */
kernfs_get_active(kn)

rdtgroup_kn_lock_live

    atomic_inc(&rdtgrp->waitcount)
    /* "c1", kn->active-- */
    kernfs_break_active_protection(kn)
    mutex_lock

rdtgroup_rmdir_ctrl

    free_all_child_rdtgrp
      sentry->flags = RDT_DELETED

    rdtgroup_ctrl_remove
      rdtgrp->flags = RDT_DELETED
      kernfs_get(kn)
      kernfs_remove(rdtgrp->kn)
        __kernfs_remove
          /* "mon_groups", sub_kn */
          atomic_add(KN_DEACTIVATED_BIAS, &sub_kn->active)
          kernfs_drain(sub_kn)
            /*
             * sub_kn->active == KN_DEACTIVATED_BIAS + 1,
             * waiting on sub_kn->active to drop, but it
             * never drops in Thread 2 which is blocked
             * on getting rdtgroup_mutex.
             */

Thread 1 hangs here ---->
wait_event(sub_kn->active == KN_DEACTIVATED_BIAS)
...
rdtgroup_mkdir
rdtgroup_mkdir_mon(parent_kn, prgrp_kn)
mkdir_rdt_prepare(parent_kn, prgrp_kn)
rdtgroup_kn_lock_live(prgrp_kn)
atomic_inc(&rdtgrp->waitcount) /*
* "c1", prgrp_kn->active-- *
* The active reference on "c1" is
* dropped, but not matching the
* actual active reference taken
* on "mon_groups", thus causing
* Thread 1 to wait forever while
* holding rdtgroup_mutex. */
kernfs_break_active_protection(
prgrp_kn)
/*
* Trying to get rdtgroup_mutex
* which is held by Thread 1. */
Thread 2 hangs here ----> mutex_lock

                                      ...

The problem is that the creation of a subdirectory in the "mon_groups" directory incorrectly releases the active protection of its parent directory instead of itself before it starts waiting for rdtgroup_mutex. This is triggered by the rdtgroup_mkdir() flow calling rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() with kernfs node of the parent control group ("c1") as argument. It should be called with kernfs node "mon_groups" instead. What is currently missing is that the kn->priv of "mon_groups" is NULL instead of pointing to the rdtgrp.

Fix it by pointing kn->priv to rdtgrp when "mon_groups" is created. Then it could be passed to rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() instead. And then it operates on the same rdtgroup structure but handles the active reference of kernfs node "mon_groups" to prevent deadlock. The same changes are also made to the "mon_data" directories.

This results in some unused function parameters that will be cleaned up in follow-up patch as the focus here is on the fix only in support of backporting efforts.

Backporting notes:

Since upstream commit fa7d949337cc ("x86/resctrl: Rename and move rdt files to a separate directory"), the file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c has been renamed and moved to arch/x86/kernel/cpu/resctrl/rdtgroup.c.
Apply the change against file arch/x86/kernel/cpu/intel_rdt_rdtgroup.c for older stable trees.

---
 arch/x86/kernel/cpu/intel_rdt_rdtgroup.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
arch/x86/kernel/cpu/intel_rdt_rdtgroup.c +8 -8
index 77770ca..11c5acc 100644
@@ -2005,7 +2005,7 @@ static struct dentry *rdt_mount(struct file_system_type *fs_type,
 
 	if (rdt_mon_capable) {
 		ret = mongroup_create_dir(rdtgroup_default.kn,
-					  NULL, "mon_groups",
+					  &rdtgroup_default, "mon_groups",
 					  &kn_mongrp);
 		if (ret) {
 			dentry = ERR_PTR(ret);
@@ -2415,7 +2415,7 @@ static int mkdir_mondata_all(struct kernfs_node *parent_kn,
 	/*
 	 * Create the mon_data directory first.
 	 */
-	ret = mongroup_create_dir(parent_kn, NULL, "mon_data", &kn);
+	ret = mongroup_create_dir(parent_kn, prgrp, "mon_data", &kn);
 	if (ret)
 		return ret;
 
@@ -2568,7 +2568,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 	uint files = 0;
 	int ret;
 
-	prdtgrp = rdtgroup_kn_lock_live(prgrp_kn);
+	prdtgrp = rdtgroup_kn_lock_live(parent_kn);
 	rdt_last_cmd_clear();
 	if (!prdtgrp) {
 		ret = -ENODEV;
@@ -2643,7 +2643,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 	kernfs_activate(kn);
 
 	/*
-	 * The caller unlocks the prgrp_kn upon success.
+	 * The caller unlocks the parent_kn upon success.
 	 */
 	return 0;
 
@@ -2654,7 +2654,7 @@ static int mkdir_rdt_prepare(struct kernfs_node *parent_kn,
 out_free_rgrp:
 	kfree(rdtgrp);
 out_unlock:
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
@@ -2692,7 +2692,7 @@ static int rdtgroup_mkdir_mon(struct kernfs_node *parent_kn,
 	 */
 	list_add_tail(&rdtgrp->mon.crdtgrp_list, &prgrp->mon.crdtgrp_list);
 
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
@@ -2735,7 +2735,7 @@ static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
 		 * Create an empty mon_groups directory to hold the subset
 		 * of tasks and cpus to monitor.
 		 */
-		ret = mongroup_create_dir(kn, NULL, "mon_groups", NULL);
+		ret = mongroup_create_dir(kn, rdtgrp, "mon_groups", NULL);
 		if (ret) {
 			rdt_last_cmd_puts("kernfs subdir error\n");
 			goto out_del_list;
@@ -2751,7 +2751,7 @@ static int rdtgroup_mkdir_ctrl_mon(struct kernfs_node *parent_kn,
 out_common_fail:
 	mkdir_rdt_prepare_clean(rdtgrp);
 out_unlock:
-	rdtgroup_kn_unlock(prgrp_kn);
+	rdtgroup_kn_unlock(parent_kn);
 	return ret;
 }
 
Signature · 1 line
1.8.3.1