coresight: tmc-etf: Fix global-out-of-bounds in tmc_update_etf_buffer()

Message ID 20210505093430.18445-1-saiprakash.ranjan@codeaurora.org
State New
Headers show
Series
  • coresight: tmc-etf: Fix global-out-of-bounds in tmc_update_etf_buffer()
Related show

Commit Message

Sai Prakash Ranjan May 5, 2021, 9:34 a.m.
commit 6f755e85c332 ("coresight: Add helper for inserting synchronization
packets") removed trailing '\0' from barrier_pkt array and updated the
call sites like etb_update_buffer() to have proper checks for barrier_pkt
size before read but missed updating tmc_update_etf_buffer() which still
reads barrier_pkt past the array size resulting in KASAN out-of-bounds
bug. Fix this by adding a check for barrier_pkt size before accessing
like it is done in etb_update_buffer().

 BUG: KASAN: global-out-of-bounds in tmc_update_etf_buffer+0x4b8/0x698
 Read of size 4 at addr ffffffd05b7d1030 by task perf/2629

 Call trace:
  dump_backtrace+0x0/0x27c
  show_stack+0x20/0x2c
  dump_stack+0x11c/0x188
  print_address_description+0x3c/0x4a4
  __kasan_report+0x140/0x164
  kasan_report+0x10/0x18
  __asan_report_load4_noabort+0x1c/0x24
  tmc_update_etf_buffer+0x4b8/0x698
  etm_event_stop+0x248/0x2d8
  etm_event_del+0x20/0x2c
  event_sched_out+0x214/0x6f0
  group_sched_out+0xd0/0x270
  ctx_sched_out+0x2ec/0x518
  __perf_event_task_sched_out+0x4fc/0xe6c
  __schedule+0x1094/0x16a0
  preempt_schedule_irq+0x88/0x170
  arm64_preempt_schedule_irq+0xf0/0x18c
  el1_irq+0xe8/0x180
  perf_event_exec+0x4d8/0x56c
  setup_new_exec+0x204/0x400
  load_elf_binary+0x72c/0x18c0
  search_binary_handler+0x13c/0x420
  load_script+0x500/0x6c4
  search_binary_handler+0x13c/0x420
  exec_binprm+0x118/0x654
  __do_execve_file+0x77c/0xba4
  __arm64_compat_sys_execve+0x98/0xac
  el0_svc_common+0x1f8/0x5e0
  el0_svc_compat_handler+0x84/0xb0
  el0_svc_compat+0x10/0x50

 The buggy address belongs to the variable:
  barrier_pkt+0x10/0x40

 Memory state around the buggy address:
  ffffffd05b7d0f00: fa fa fa fa 04 fa fa fa fa fa fa fa 00 00 00 00
  ffffffd05b7d0f80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 >ffffffd05b7d1000: 00 00 00 00 00 00 fa fa fa fa fa fa 00 00 00 03
                                      ^
  ffffffd05b7d1080: fa fa fa fa 00 02 fa fa fa fa fa fa 03 fa fa fa
  ffffffd05b7d1100: fa fa fa fa 00 00 00 00 05 fa fa fa fa fa fa fa
 ==================================================================

Fixes: 6f755e85c332 ("coresight: Add helper for inserting synchronization packets")
Cc: stable@vger.kernel.org
Signed-off-by: Sai Prakash Ranjan <saiprakash.ranjan@codeaurora.org>
---
 drivers/hwtracing/coresight/coresight-tmc-etf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Suzuki K Poulose May 25, 2021, 8:54 a.m. | #1
Hi Sai

On 05/05/2021 10:47, Suzuki K Poulose wrote:
> On 05/05/2021 10:34, Sai Prakash Ranjan wrote:

>> commit 6f755e85c332 ("coresight: Add helper for inserting synchronization

>> packets") removed trailing '\0' from barrier_pkt array and updated the

>> call sites like etb_update_buffer() to have proper checks for barrier_pkt

>> size before read but missed updating tmc_update_etf_buffer() which still

>> reads barrier_pkt past the array size resulting in KASAN out-of-bounds

>> bug. Fix this by adding a check for barrier_pkt size before accessing

>> like it is done in etb_update_buffer().

>>

>>   BUG: KASAN: global-out-of-bounds in tmc_update_etf_buffer+0x4b8/0x698

>>   Read of size 4 at addr ffffffd05b7d1030 by task perf/2629

>>

>>   Call trace:

>>    dump_backtrace+0x0/0x27c

>>    show_stack+0x20/0x2c

>>    dump_stack+0x11c/0x188

>>    print_address_description+0x3c/0x4a4

>>    __kasan_report+0x140/0x164

>>    kasan_report+0x10/0x18

>>    __asan_report_load4_noabort+0x1c/0x24

>>    tmc_update_etf_buffer+0x4b8/0x698

>>    etm_event_stop+0x248/0x2d8

>>    etm_event_del+0x20/0x2c

>>    event_sched_out+0x214/0x6f0

>>    group_sched_out+0xd0/0x270

>>    ctx_sched_out+0x2ec/0x518

>>    __perf_event_task_sched_out+0x4fc/0xe6c

>>    __schedule+0x1094/0x16a0

>>    preempt_schedule_irq+0x88/0x170

>>    arm64_preempt_schedule_irq+0xf0/0x18c

>>    el1_irq+0xe8/0x180

>>    perf_event_exec+0x4d8/0x56c

>>    setup_new_exec+0x204/0x400

>>    load_elf_binary+0x72c/0x18c0

>>    search_binary_handler+0x13c/0x420

>>    load_script+0x500/0x6c4

>>    search_binary_handler+0x13c/0x420

>>    exec_binprm+0x118/0x654

>>    __do_execve_file+0x77c/0xba4

>>    __arm64_compat_sys_execve+0x98/0xac

>>    el0_svc_common+0x1f8/0x5e0

>>    el0_svc_compat_handler+0x84/0xb0

>>    el0_svc_compat+0x10/0x50

>>

>>   The buggy address belongs to the variable:

>>    barrier_pkt+0x10/0x40

>>

>>   Memory state around the buggy address:

>>    ffffffd05b7d0f00: fa fa fa fa 04 fa fa fa fa fa fa fa 00 00 00 00

>>    ffffffd05b7d0f80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

>>   >ffffffd05b7d1000: 00 00 00 00 00 00 fa fa fa fa fa fa 00 00 00 03

>>                                        ^

>>    ffffffd05b7d1080: fa fa fa fa 00 02 fa fa fa fa fa fa 03 fa fa fa

>>    ffffffd05b7d1100: fa fa fa fa 00 00 00 00 05 fa fa fa fa fa fa fa

>>   ==================================================================

>>

>> Fixes: 6f755e85c332 ("coresight: Add helper for inserting 

>> synchronization packets")


I have changed the commit to :

Fixes: 0c3fc4d5fa26 ("coresight: Add barrier packet for synchronisation")

Applied.

Thanks
Suzuki
Sai Prakash Ranjan May 25, 2021, 11:47 a.m. | #2
On 2021-05-25 14:24, Suzuki K Poulose wrote:
> Hi Sai

> 

> On 05/05/2021 10:47, Suzuki K Poulose wrote:

>> On 05/05/2021 10:34, Sai Prakash Ranjan wrote:

>>> commit 6f755e85c332 ("coresight: Add helper for inserting 

>>> synchronization

>>> packets") removed trailing '\0' from barrier_pkt array and updated 

>>> the

>>> call sites like etb_update_buffer() to have proper checks for 

>>> barrier_pkt

>>> size before read but missed updating tmc_update_etf_buffer() which 

>>> still

>>> reads barrier_pkt past the array size resulting in KASAN 

>>> out-of-bounds

>>> bug. Fix this by adding a check for barrier_pkt size before accessing

>>> like it is done in etb_update_buffer().

>>> 

>>>   BUG: KASAN: global-out-of-bounds in 

>>> tmc_update_etf_buffer+0x4b8/0x698

>>>   Read of size 4 at addr ffffffd05b7d1030 by task perf/2629

>>> 

>>>   Call trace:

>>>    dump_backtrace+0x0/0x27c

>>>    show_stack+0x20/0x2c

>>>    dump_stack+0x11c/0x188

>>>    print_address_description+0x3c/0x4a4

>>>    __kasan_report+0x140/0x164

>>>    kasan_report+0x10/0x18

>>>    __asan_report_load4_noabort+0x1c/0x24

>>>    tmc_update_etf_buffer+0x4b8/0x698

>>>    etm_event_stop+0x248/0x2d8

>>>    etm_event_del+0x20/0x2c

>>>    event_sched_out+0x214/0x6f0

>>>    group_sched_out+0xd0/0x270

>>>    ctx_sched_out+0x2ec/0x518

>>>    __perf_event_task_sched_out+0x4fc/0xe6c

>>>    __schedule+0x1094/0x16a0

>>>    preempt_schedule_irq+0x88/0x170

>>>    arm64_preempt_schedule_irq+0xf0/0x18c

>>>    el1_irq+0xe8/0x180

>>>    perf_event_exec+0x4d8/0x56c

>>>    setup_new_exec+0x204/0x400

>>>    load_elf_binary+0x72c/0x18c0

>>>    search_binary_handler+0x13c/0x420

>>>    load_script+0x500/0x6c4

>>>    search_binary_handler+0x13c/0x420

>>>    exec_binprm+0x118/0x654

>>>    __do_execve_file+0x77c/0xba4

>>>    __arm64_compat_sys_execve+0x98/0xac

>>>    el0_svc_common+0x1f8/0x5e0

>>>    el0_svc_compat_handler+0x84/0xb0

>>>    el0_svc_compat+0x10/0x50

>>> 

>>>   The buggy address belongs to the variable:

>>>    barrier_pkt+0x10/0x40

>>> 

>>>   Memory state around the buggy address:

>>>    ffffffd05b7d0f00: fa fa fa fa 04 fa fa fa fa fa fa fa 00 00 00 00

>>>    ffffffd05b7d0f80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

>>>   >ffffffd05b7d1000: 00 00 00 00 00 00 fa fa fa fa fa fa 00 00 00 03

>>>                                        ^

>>>    ffffffd05b7d1080: fa fa fa fa 00 02 fa fa fa fa fa fa 03 fa fa fa

>>>    ffffffd05b7d1100: fa fa fa fa 00 00 00 00 05 fa fa fa fa fa fa fa

>>>   ==================================================================

>>> 

>>> Fixes: 6f755e85c332 ("coresight: Add helper for inserting 

>>> synchronization packets")

> 

> I have changed the commit to :

> 

> Fixes: 0c3fc4d5fa26 ("coresight: Add barrier packet for 

> synchronisation")

> 

> Applied.

> 


Sure, thanks Suzuki.

Regards,
Sai

-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a 
member
of Code Aurora Forum, hosted by The Linux Foundation

Patch

diff --git a/drivers/hwtracing/coresight/coresight-tmc-etf.c b/drivers/hwtracing/coresight/coresight-tmc-etf.c
index 45b85edfc690..cd0fb7bfba68 100644
--- a/drivers/hwtracing/coresight/coresight-tmc-etf.c
+++ b/drivers/hwtracing/coresight/coresight-tmc-etf.c
@@ -530,7 +530,7 @@  static unsigned long tmc_update_etf_buffer(struct coresight_device *csdev,
 		buf_ptr = buf->data_pages[cur] + offset;
 		*buf_ptr = readl_relaxed(drvdata->base + TMC_RRD);
 
-		if (lost && *barrier) {
+		if (lost && i < CORESIGHT_BARRIER_PKT_SIZE) {
 			*buf_ptr = *barrier;
 			barrier++;
 		}