atm: Fix use-after-free bug in atm_dev_register()
Subsystems (3)
PATCH 1/1 atm: Fix use-after-free bug in atm_dev_register()
When device_register() return failed in atm_register_sysfs, the program will return to atm_dev_register and will kfree the device. As the comment of device_register() says, put_device() needs to be used to give up the reference in the error path. Using kfree instead triggers a UAF, as shown by the following KASAN report, obtained by causing device_register() to fail. This patch calls put_device instead of kfree when atm_register_sysfs has failed, and call kfree only when atm_proc_dev_register has failed.
KASAN report details as below:
[ 94.341664] BUG: KASAN: use-after-free in sysfs_kf_seq_show+0x306/0x440 [ 94.341674] Read of size 8 at addr ffff88819a8a30e8 by task systemd-udevd/484
[ 94.341680] CPU: 3 PID: 484 Comm: systemd-udevd Tainted: G E 6.1.0-rc1+ #1 [ 94.341684] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 02/27/2020 [ 94.341703] Call Trace: [ 94.341705] <TASK> [ 94.341707] dump_stack_lvl+0x49/0x63 [ 94.341713] print_report+0x177/0x46e [ 94.341717] ? kasan_complete_mode_report_info+0x7c/0x210 [ 94.341720] ? sysfs_kf_seq_show+0x306/0x440 [ 94.341753] kasan_report+0xb0/0x140 [ 94.341757] ? sysfs_kf_seq_show+0x306/0x440 [ 94.341760] __asan_report_load8_noabort+0x14/0x20 [ 94.341763] sysfs_kf_seq_show+0x306/0x440 [ 94.341766] kernfs_seq_show+0x145/0x1b0 [ 94.341769] seq_read_iter+0x408/0x1080 [ 94.341774] kernfs_fop_read_iter+0x3d5/0x540 [ 94.341794] vfs_read+0x542/0x800 [ 94.341797] ? kernel_read+0x130/0x130 [ 94.341800] ? __kasan_check_read+0x11/0x20 [ 94.341824] ? get_nth_filter.part.0+0x200/0x200 [ 94.341828] ksys_read+0x116/0x220 [ 94.341831] ? __ia32_sys_pwrite64+0x1f0/0x1f0 [ 94.341849] ? __secure_computing+0x17c/0x2d0 [ 94.341852] __x64_sys_read+0x72/0xb0 [ 94.341875] do_syscall_64+0x59/0x90 [ 94.341878] ? exc_page_fault+0x72/0xf0 [ 94.341881] entry_SYSCALL_64_after_hwframe+0x63/0xcd [ 94.341885] RIP: 0033:0x7fc391f14992 [ 94.341888] Code: c0 e9 b2 fe ff ff 50 48 8d 3d fa b2 0c 00 e8 c5 1d 02 00 0f 1f 44 00 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 0f 05 <48> 3d 00 f0 ff ff 77 56 c3 0f 1f 44 00 00 48 83 ec 28 48 89 54 24 [ 94.341891] RSP: 002b:00007ffe33fed818 EFLAGS: 00000246 ORIG_RAX: 0000000000000000 [ 94.341896] RAX: ffffffffffffffda RBX: 0000000000001018 RCX: 00007fc391f14992 [ 94.341898] RDX: 0000000000001018 RSI: 0000558a696b0880 RDI: 000000000000000e [ 94.341900] RBP: 0000558a696b0880 R08: 0000000000000000 R09: 0000558a696b0880 [ 94.341902] R10: 00007fc39201a300 R11: 0000000000000246 R12: 000000000000000e [ 94.341904] R13: 0000000000001017 R14: 0000000000000002 R15: 00007ffe33fed840 [ 94.341908] </TASK>
[ 94.341911] Allocated by task 2613: [ 94.341914] kasan_save_stack+0x26/0x50 [ 94.341932] kasan_set_track+0x25/0x40 [ 94.341934] kasan_save_alloc_info+0x1e/0x30 [ 94.341936] __kasan_kmalloc+0xb4/0xc0 [ 94.341938] kmalloc_trace+0x4a/0xb0 [ 94.341941] atm_dev_register+0x5d/0x700 [atm] [ 94.341949] atmtcp_create+0x77/0x1f0 [atmtcp] [ 94.341953] atmtcp_ioctl+0x12d/0xb9f [atmtcp] [ 94.341957] do_vcc_ioctl+0xfe/0x640 [atm] [ 94.341962] vcc_ioctl+0x10/0x20 [atm] [ 94.341968] svc_ioctl+0x587/0x6c0 [atm] [ 94.341973] sock_do_ioctl+0xd7/0x1e0 [ 94.341977] sock_ioctl+0x1b5/0x560 [ 94.341979] __x64_sys_ioctl+0x132/0x1b0 [ 94.341981] do_syscall_64+0x59/0x90 [ 94.341983] entry_SYSCALL_64_after_hwframe+0x63/0xcd
[ 94.341986] Freed by task 2613: [ 94.341988] kasan_save_stack+0x26/0x50 [ 94.341991] kasan_set_track+0x25/0x40 [ 94.341993] kasan_save_free_info+0x2e/0x50 [ 94.341995] ____kasan_slab_free+0x174/0x1e0 [ 94.341997] __kasan_slab_free+0x12/0x20 [ 94.342000] slab_free_freelist_hook+0xd0/0x1a0 [ 94.342002] __kmem_cache_free+0x193/0x2c0 [ 94.342005] kfree+0x79/0x120 [ 94.342007] atm_dev_register.cold+0x46/0x64 [atm] [ 94.342013] atmtcp_create+0x77/0x1f0 [atmtcp] [ 94.342016] atmtcp_ioctl+0x12d/0xb9f [atmtcp] [ 94.342020] do_vcc_ioctl+0xfe/0x640 [atm] [ 94.342077] vcc_ioctl+0x10/0x20 [atm] [ 94.342083] svc_ioctl+0x587/0x6c0 [atm] [ 94.342088] sock_do_ioctl+0xd7/0x1e0 [ 94.342091] sock_ioctl+0x1b5/0x560 [ 94.342093] __x64_sys_ioctl+0x132/0x1b0 [ 94.342095] do_syscall_64+0x59/0x90 [ 94.342098] entry_SYSCALL_64_after_hwframe+0x63/0xcd
[ 94.342102] The buggy address belongs to the object at ffff88819a8a3000 which belongs to the cache kmalloc-1k of size 1024 [ 94.342105] The buggy address is located 232 bytes inside of 1024-byte region [ffff88819a8a3000, ffff88819a8a3400)
[ 94.342109] The buggy address belongs to the physical page: [ 94.342111] page:0000000099993f0a refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x19a8a0 [ 94.342114] head:0000000099993f0a order:3 compound_mapcount:0 compound_pincount:0 [ 94.342116] flags: 0x17ffffc0010200(slab|head|node=0|zone=2|lastcpupid=0x1fffff) [ 94.342136] raw: 0017ffffc0010200 dead000000000100 dead000000000122 ffff888100042dc0 [ 94.342138] raw: 0000000000000000 0000000080100010 00000001ffffffff 0000000000000000 [ 94.342139] page dumped because: kasan: bad access detected
[ 94.342141] Memory state around the buggy address: [ 94.342143] ffff88819a8a2f80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc [ 94.342145] ffff88819a8a3000: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 94.342147] >ffff88819a8a3080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 94.342148] ^ [ 94.342150] ffff88819a8a3100: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 94.342152] ffff88819a8a3180: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
--- net/atm/resources.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
net/atm/resources.c +2 -1
@@ -112,12 +112,14 @@ struct atm_dev *atm_dev_register(const char *type, struct device *parent,if (atm_proc_dev_register (dev) <0 ) {pr_err ("atm_proc_dev_register failed for dev %s\n" , type); +kfree (dev); goto out_fail; }if (atm_register_sysfs (dev, parent) <0 ) {pr_err ("atm_register_sysfs failed for dev %s\n" , type);atm_proc_dev_deregister (dev); +put_device (&dev->class_dev); goto out_fail; }
@@ -128,7 +130,6 @@ struct atm_dev *atm_dev_register(const char *type, struct device *parent,return dev; out_fail: -kfree (dev); dev =NULL ; goto out; }
Signature · 1 line
2.34.1
Quoted 109 lines · Nir Levy
When
device_register()return failed inatm_register_sysfs, the program will return toatm_dev_registerand will kfree the device. As the comment ofdevice_register()says,put_device()needs to be used to give up the reference in the error path. Using kfree instead triggers a UAF, as shown by the following KASAN report, obtained by causingdevice_register()to fail. This patch callsput_deviceinstead of kfree whenatm_register_sysfshas failed, and call kfree only whenatm_proc_dev_registerhas failed.KASAN report details as below:
[ 94.341664] BUG: KASAN: use-after-free in sysfs_kf_seq_show+0x306/0x440 [ 94.341674] Read of size 8 at addr ffff88819a8a30e8 by task systemd-udevd/484[ 94.341680] CPU: 3 PID: 484 Comm: systemd-udevd Tainted: G E 6.1.0-rc1+ #1 [ 94.341684] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 02/27/2020 [ 94.341703] Call Trace: [ 94.341705] <TASK> [ 94.341707] dump_stack_lvl+0x49/0x63 [ 94.341713] print_report+0x177/0x46e [ 94.341717] ? kasan_complete_mode_report_info+0x7c/0x210 [ 94.341720] ? sysfs_kf_seq_show+0x306/0x440 [ 94.341753] kasan_report+0xb0/0x140 [ 94.341757] ? sysfs_kf_seq_show+0x306/0x440 [ 94.341760] __asan_report_load8_noabort+0x14/0x20 [ 94.341763] sysfs_kf_seq_show+0x306/0x440 [ 94.341766] kernfs_seq_show+0x145/0x1b0 [ 94.341769] seq_read_iter+0x408/0x1080 [ 94.341774] kernfs_fop_read_iter+0x3d5/0x540 [ 94.341794] vfs_read+0x542/0x800 [ 94.341797] ? kernel_read+0x130/0x130 [ 94.341800] ? __kasan_check_read+0x11/0x20 [ 94.341824] ? get_nth_filter.part.0+0x200/0x200 [ 94.341828] ksys_read+0x116/0x220 [ 94.341831] ? __ia32_sys_pwrite64+0x1f0/0x1f0 [ 94.341849] ? __secure_computing+0x17c/0x2d0 [ 94.341852] __x64_sys_read+0x72/0xb0 [ 94.341875] do_syscall_64+0x59/0x90 [ 94.341878] ? exc_page_fault+0x72/0xf0 [ 94.341881] entry_SYSCALL_64_after_hwframe+0x63/0xcd [ 94.341885] RIP: 0033:0x7fc391f14992 [ 94.341888] Code: c0 e9 b2 fe ff ff 50 48 8d 3d fa b2 0c 00 e8 c5 1d 02 00 0f 1f 44 00 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 0f 05 <48> 3d 00 f0 ff ff 77 56 c3 0f 1f 44 00 00 48 83 ec 28 48 89 54 24 [ 94.341891] RSP: 002b:00007ffe33fed818 EFLAGS: 00000246 ORIG_RAX: 0000000000000000 [ 94.341896] RAX: ffffffffffffffda RBX: 0000000000001018 RCX: 00007fc391f14992 [ 94.341898] RDX: 0000000000001018 RSI: 0000558a696b0880 RDI: 000000000000000e [ 94.341900] RBP: 0000558a696b0880 R08: 0000000000000000 R09: 0000558a696b0880 [ 94.341902] R10: 00007fc39201a300 R11: 0000000000000246 R12: 000000000000000e [ 94.341904] R13: 0000000000001017 R14: 0000000000000002 R15: 00007ffe33fed840 [ 94.341908] </TASK>[ 94.341911] Allocated by task 2613: [ 94.341914] kasan_save_stack+0x26/0x50 [ 94.341932] kasan_set_track+0x25/0x40 [ 94.341934] kasan_save_alloc_info+0x1e/0x30 [ 94.341936] __kasan_kmalloc+0xb4/0xc0 [ 94.341938] kmalloc_trace+0x4a/0xb0 [ 94.341941] atm_dev_register+0x5d/0x700 [atm] [ 94.341949] atmtcp_create+0x77/0x1f0 [atmtcp] [ 94.341953] atmtcp_ioctl+0x12d/0xb9f [atmtcp] [ 94.341957] do_vcc_ioctl+0xfe/0x640 [atm] [ 94.341962] vcc_ioctl+0x10/0x20 [atm] [ 94.341968] svc_ioctl+0x587/0x6c0 [atm] [ 94.341973] sock_do_ioctl+0xd7/0x1e0 [ 94.341977] sock_ioctl+0x1b5/0x560 [ 94.341979] __x64_sys_ioctl+0x132/0x1b0 [ 94.341981] do_syscall_64+0x59/0x90 [ 94.341983] entry_SYSCALL_64_after_hwframe+0x63/0xcd[ 94.341986] Freed by task 2613: [ 94.341988] kasan_save_stack+0x26/0x50 [ 94.341991] kasan_set_track+0x25/0x40 [ 94.341993] kasan_save_free_info+0x2e/0x50 [ 94.341995] ____kasan_slab_free+0x174/0x1e0 [ 94.341997] __kasan_slab_free+0x12/0x20 [ 94.342000] slab_free_freelist_hook+0xd0/0x1a0 [ 94.342002] __kmem_cache_free+0x193/0x2c0 [ 94.342005] kfree+0x79/0x120 [ 94.342007] atm_dev_register.cold+0x46/0x64 [atm] [ 94.342013] atmtcp_create+0x77/0x1f0 [atmtcp] [ 94.342016] atmtcp_ioctl+0x12d/0xb9f [atmtcp] [ 94.342020] do_vcc_ioctl+0xfe/0x640 [atm] [ 94.342077] vcc_ioctl+0x10/0x20 [atm] [ 94.342083] svc_ioctl+0x587/0x6c0 [atm] [ 94.342088] sock_do_ioctl+0xd7/0x1e0 [ 94.342091] sock_ioctl+0x1b5/0x560 [ 94.342093] __x64_sys_ioctl+0x132/0x1b0 [ 94.342095] do_syscall_64+0x59/0x90 [ 94.342098] entry_SYSCALL_64_after_hwframe+0x63/0xcd[ 94.342102] The buggy address belongs to the object at ffff88819a8a3000 which belongs to the cache kmalloc-1k of size 1024 [ 94.342105] The buggy address is located 232 bytes inside of 1024-byte region [ffff88819a8a3000, ffff88819a8a3400)[ 94.342109] The buggy address belongs to the physical page: [ 94.342111] page:0000000099993f0a refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x19a8a0 [ 94.342114] head:0000000099993f0a order:3 compound_mapcount:0 compound_pincount:0 [ 94.342116] flags: 0x17ffffc0010200(slab|head|node=0|zone=2|lastcpupid=0x1fffff) [ 94.342136] raw: 0017ffffc0010200 dead000000000100 dead000000000122 ffff888100042dc0 [ 94.342138] raw: 0000000000000000 0000000080100010 00000001ffffffff 0000000000000000 [ 94.342139] page dumped because: kasan: bad access detected[ 94.342141] Memory state around the buggy address: [ 94.342143] ffff88819a8a2f80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc [ 94.342145] ffff88819a8a3000: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 94.342147] >ffff88819a8a3080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 94.342148] ^ [ 94.342150] ffff88819a8a3100: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [ 94.342152] ffff88819a8a3180: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
Quoted 4 lines
Signed-off-by Nir Levy <[email protected]>--- net/atm/resources.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
Please add target in patch title - {PATCH net] ... There is a need to add Fixes line too.
Quoted 7 lines
net/atm/resources.c +0
-0@@ -112,12 +112,14 @@ struct atm_dev *atm_dev_register(const char *type, struct device *parent,
net/atm/resources.c struct atm_dev *atm_dev_register(const char *type, struct device *parent, · line 122
Quoted 10 lines
net/atm/resources.c +2
-0if (atm_proc_dev_register (dev) <0 ) {pr_err ("atm_proc_dev_register failed for dev %s\n" , type); +kfree (dev); goto out_fail; }if (atm_register_sysfs (dev, parent) <0 ) {pr_err ("atm_register_sysfs failed for dev %s\n" , type);atm_proc_dev_deregister (dev); +put_device (&dev->class_dev);
The right fix is to change atm_register_sysfs() to call this put_device and worth to get rid from device_del() which should be replaced with device_unregister().
Thanks
Quoted 14 lines
net/atm/resources.c +0
-1goto out_fail; }@@ -128,7 +130,6 @@ struct atm_dev *atm_dev_register(const char *type, struct device *parent,return dev; out_fail: -kfree (dev); dev =NULL ; goto out; }Signature · 1 line
2.34.1
Jakub Kicinski
maintainerQuoted 5 lines · On Tue, 6 Dec 2022 00:08:32 +0200 Nir Levy wrote:
From
5cb38a02b10a2f52bf7f7eef67fa4abc0974b21bMon Sep 17 00:00:00 2001
From: Nir Levy <[email protected]>
Date: Fri, 25 Nov 2022 21:06:48 +0200
Subject: [PATCH net] net: atm: Fix use-after-free bug inatm_dev_register()
Quoted 9 lines
When
device_register()return failed inatm_register_sysfs, the program will return toatm_dev_registerand will kfree the device. As the comment ofdevice_register()says,put_device()needs to be used to give up the reference in the error path. Using kfree instead triggers a UAF, as shown by the following KASAN report, obtained by causingdevice_register()to fail. This patch callsput_devicewhenatm_register_sysfshas failed, and call kfree only whenatm_proc_dev_registerhas failed.
Please make a fresh submission, it doesn't have to be in the same thread (in fact that's discouraged).