Skip to contentSkip to conversation

pci: endpoint test BUG

DISCUSSION Randy Dunlap 3 participants 7 messages Sep 16, 2019 – Sep 23, 2019 2 messages not in the archive PCI

lore
Share & tools
lore

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

Randy Dunlap

author Tested-by
#1 lore raw
Quoted 24 lines · Hillf Danton

On Sun, 15 Sep 2019 09:34:37 -0700

Quoted 11 lines

Kernel is 5.3-rc8 on x86_64.

Loading and removing the pci-epf-test module causes a BUG.

[40928.435755] calling  pci_epf_test_init+0x0/0x1000 [pci_epf_test] @ 12132
[40928.436717] initcall pci_epf_test_init+0x0/0x1000 [pci_epf_test] returned 0 after 891 usecs
[40936.996081] ==================================================================
[40936.996125] BUG: KASAN: use-after-free in pci_epf_remove_cfs+0x1ae/0x1f0
[40936.996153] Write of size 8 at addr ffff88810a22a068 by task rmmod/12139

Fix fb0de5b8dcc6 and ef1433f717a2 if the current group::group_entry used by pci epf does not break how configfs uses it.

drivers/pci/endpoint/pci-epf-core.c +0 -0
@@ -153,9 +153,11 @@ static void pci_epf_remove_cfs(struct pc
 		return;
 
 	mutex_lock(&pci_epf_mutex);

drivers/pci/endpoint/pci-epf-core.c static void pci_epf_remove_cfs(struct pc · line 163

Quoted 10 lines
drivers/pci/endpoint/pci-epf-core.c +4 -2
-	list_for_each_entry_safe(group, tmp, &driver->epf_group, group_entry)
+	list_for_each_entry_safe(group, tmp, &driver->epf_group,
+							group_entry) {
+		list_del_init(&group->group_entry);
 		pci_ep_cfs_remove_epf_group(group);
-	list_del(&driver->epf_group);
+	}
 	mutex_unlock(&pci_epf_mutex);
 }
 

Fixes the problem for me. Thanks.

Please make a proper patch.

Signature · 1 line
~Randy
#2 lore raw
In reply to a message not in the archive
Quoted 22 lines · Hillf Danton

On Sun, 15 Sep 2019 09:34:37 -0700

Quoted 11 lines

Kernel is 5.3-rc8 on x86_64.

Loading and removing the pci-epf-test module causes a BUG.

[40928.435755] calling  pci_epf_test_init+0x0/0x1000 [pci_epf_test] @ 12132
[40928.436717] initcall pci_epf_test_init+0x0/0x1000 [pci_epf_test] returned 0 after 891 usecs
[40936.996081] ==================================================================
[40936.996125] BUG: KASAN: use-after-free in pci_epf_remove_cfs+0x1ae/0x1f0
[40936.996153] Write of size 8 at addr ffff88810a22a068 by task rmmod/12139

Fix fb0de5b8dcc6 and ef1433f717a2 if the current group::group_entry used by pci epf does not break how configfs uses it.

drivers/pci/endpoint/pci-epf-core.c +0 -0
@@ -153,9 +153,11 @@ static void pci_epf_remove_cfs(struct pc
 		return;
 

drivers/pci/endpoint/pci-epf-core.c static void pci_epf_remove_cfs(struct pc · line 162

Quoted 10 lines
drivers/pci/endpoint/pci-epf-core.c +4 -2
 	mutex_lock(&pci_epf_mutex);
-	list_for_each_entry_safe(group, tmp, &driver->epf_group, group_entry)
+	list_for_each_entry_safe(group, tmp, &driver->epf_group,
+							group_entry) {
+		list_del_init(&group->group_entry);
 		pci_ep_cfs_remove_epf_group(group);
-	list_del(&driver->epf_group);
+	}
 	mutex_unlock(&pci_epf_mutex);
 }

Thank you Hillf. Kishon, can you confirm that's the proper fix for this bug please ? I would like to turn this into a patch and merge it in the upcoming merge window PR so it ought to be fairly quick, please let me know asap.

Lorenzo

#3 lore raw
Quoted 23 lines · Lorenzo Pieralisi
Quoted 22 lines · Hillf Danton

On Sun, 15 Sep 2019 09:34:37 -0700

Quoted 11 lines

Kernel is 5.3-rc8 on x86_64.

Loading and removing the pci-epf-test module causes a BUG.

[40928.435755] calling  pci_epf_test_init+0x0/0x1000 [pci_epf_test] @ 12132
[40928.436717] initcall pci_epf_test_init+0x0/0x1000 [pci_epf_test] returned 0 after 891 usecs
[40936.996081] ==================================================================
[40936.996125] BUG: KASAN: use-after-free in pci_epf_remove_cfs+0x1ae/0x1f0
[40936.996153] Write of size 8 at addr ffff88810a22a068 by task rmmod/12139

Fix fb0de5b8dcc6 and ef1433f717a2 if the current group::group_entry used by pci epf does not break how configfs uses it.

drivers/pci/endpoint/pci-epf-core.c +0 -0
@@ -153,9 +153,11 @@ static void pci_epf_remove_cfs(struct pc
 		return;
 
Quoted 10 lines

drivers/pci/endpoint/pci-epf-core.c static void pci_epf_remove_cfs(struct pc · line 162

Quoted 10 lines
drivers/pci/endpoint/pci-epf-core.c +4 -2
 	mutex_lock(&pci_epf_mutex);
-	list_for_each_entry_safe(group, tmp, &driver->epf_group, group_entry)
+	list_for_each_entry_safe(group, tmp, &driver->epf_group,
+							group_entry) {
+		list_del_init(&group->group_entry);
 		pci_ep_cfs_remove_epf_group(group);
-	list_del(&driver->epf_group);
+	}
 	mutex_unlock(&pci_epf_mutex);
 }
Quoted 8 lines

Thank you Hillf. Kishon, can you confirm that's the proper fix for this bug please ? I would like to turn this into a patch and merge it in the upcoming merge window PR so it ought to be fairly quick, please let me know asap.

Lorenzo

#4 lore raw
Quoted 25 lines · Kishon Vijay Abraham I
Quoted 23 lines · Lorenzo Pieralisi
Quoted 22 lines · Hillf Danton

On Sun, 15 Sep 2019 09:34:37 -0700

Quoted 11 lines

Kernel is 5.3-rc8 on x86_64.

Loading and removing the pci-epf-test module causes a BUG.

[40928.435755] calling  pci_epf_test_init+0x0/0x1000 [pci_epf_test] @ 12132
[40928.436717] initcall pci_epf_test_init+0x0/0x1000 [pci_epf_test] returned 0 after 891 usecs
[40936.996081] ==================================================================
[40936.996125] BUG: KASAN: use-after-free in pci_epf_remove_cfs+0x1ae/0x1f0
[40936.996153] Write of size 8 at addr ffff88810a22a068 by task rmmod/12139

Fix fb0de5b8dcc6 and ef1433f717a2 if the current group::group_entry used by pci epf does not break how configfs uses it.

drivers/pci/endpoint/pci-epf-core.c +0 -0
@@ -153,9 +153,11 @@ static void pci_epf_remove_cfs(struct pc
 		return;
 
Quoted 19 lines
Quoted 10 lines

drivers/pci/endpoint/pci-epf-core.c static void pci_epf_remove_cfs(struct pc · line 162

Quoted 10 lines
drivers/pci/endpoint/pci-epf-core.c +4 -2
 	mutex_lock(&pci_epf_mutex);
-	list_for_each_entry_safe(group, tmp, &driver->epf_group, group_entry)
+	list_for_each_entry_safe(group, tmp, &driver->epf_group,
+							group_entry) {
+		list_del_init(&group->group_entry);
 		pci_ep_cfs_remove_epf_group(group);
-	list_del(&driver->epf_group);
+	}
 	mutex_unlock(&pci_epf_mutex);
 }
Quoted 5 lines

Thank you Hillf. Kishon, can you confirm that's the proper fix for this bug please ? I would like to turn this into a patch and merge it in the upcoming merge window PR so it ought to be fairly quick, please let me know asap.

Kishon, Hillf, can you turn it into a patch and send it asap please ?

Thanks,
Lorenzo

#5 lore raw
Quoted 36 lines · Hillf Danton
Quoted 1 line

Kishon, Hillf, can you turn it into a patch and send it asap please ?

What was sent a couple of days before,

To: Bjorn Helgaas <[email protected]>

Cc linux-pci <[email protected]>, LKML <[email protected]>

Subject: [PATCH] PCI: endpoint: Fix uaf on unregistering driver

...

Fixes ef1433f717a2 ("PCI: endpoint: Create configfs entry for each pci_epf_device_id table entry")
Reported-and-tested-by Randy Dunlap <[email protected]>
Cc Al Viro <[email protected]>
Cc Dan Carpenter <[email protected]>
Cc Lorenzo Pieralisi <[email protected]>
Cc Kishon Vijay Abraham I <[email protected]>
Cc Andrey Konovalov <[email protected]>

---

Quoted 7 lines

and it is certain that <[email protected]> is on the Cc list.

It will be resent if no one saw the message.

I didn't see it and I can't find it on lore.kernel.org/linux-pci/.

Signature · 1 line
~Randy
#6 lore raw
Quoted 46 lines · Hillf Danton
Quoted 9 lines
Quoted 3 lines

It will be resent if no one saw the message.

I didn't see it and I can't find it on lore.kernel.org/linux-pci/.

Respin, git send-email works/jj/pci-epf-uaf.txt

...

From: Hillf Danton <[email protected]>

To: Bjorn Helgaas <[email protected]>

Cc linux-pci <[email protected]>,

        LKML <[email protected]>,

        Randy Dunlap <[email protected]>,

        Al Viro <[email protected]>,

        Dan Carpenter <[email protected]>,

        Lorenzo Pieralisi <[email protected]>,

        Kishon Vijay Abraham I <[email protected]>,

        Andrey Konovalov <[email protected]>,

        Hillf Danton <[email protected]>

Subject: [PATCH] PCI: endpoint: Fix uaf on unregistering driver

Date: Sat, 21 Sep 2019 09:58:28 +0800

Message-Id: <[email protected]>

MIME-Version: 1.0

Content-Transfer-Encoding: 8bit

Quoted 7 lines

Result: 250

And let me know you see it.

No, not seeing the patch in my Inbox nor on lore.kernel.org.

It's a mystery to me.

Signature · 1 line
~Randy

A message that is not in the archive (sent off-list or not yet received): c2cadd96-a6d5-45f9-9abc-4c89b4a8b056@VE1EUR03FT044.eop-EUR03.prod.protection.outlook.com

#8 lore raw
Quoted 48 lines · Hillf Danton
Quoted 9 lines
Quoted 3 lines

It will be resent if no one saw the message.

I didn't see it and I can't find it on lore.kernel.org/linux-pci/.

Respin, git send-email works/jj/pci-epf-uaf.txt

...

From: Hillf Danton <[email protected]>

To: Bjorn Helgaas <[email protected]>

Cc linux-pci <[email protected]>,

LKML <[email protected]>,

Randy Dunlap <[email protected]>,

Al Viro <[email protected]>,

Dan Carpenter <[email protected]>,

Lorenzo Pieralisi <[email protected]>,

Kishon Vijay Abraham I <[email protected]>,

Andrey Konovalov <[email protected]>,

Hillf Danton <[email protected]>

Subject: [PATCH] PCI: endpoint: Fix uaf on unregistering driver

Date: Sat, 21 Sep 2019 09:58:28 +0800

Message-Id: <[email protected]>

MIME-Version: 1.0

Content-Transfer-Encoding: 8bit

Quoted 5 lines

Result: 250

And let me know you see it.

I do not think any of your messages hit the vger mailing lists, avoid html.

I have not received any patch and they are not in the mailing list archives either.

Please check your email/SMTP set-up. I do not apply patches that are not on linux-pci@vger, worst case you can attach the patch to your reply and I can send it on your behalf but we must do this quickly.

Lorenzo

Quoted 4 lines

Thanks

Hillf