tools/testing/nvdimm: Fix the index for dimm devices.
PATCH 1/1 tools/testing/nvdimm: Fix the index for dimm devices.
From: Masayoshi Mizuma <[email protected]>
KASAN reports following global out of bounds access while nfit_test is being loaded. The out of bound access happens the following reference to dimm_fail_cmd_flags[dimm]. 'dimm' is over than the index value, NUM_DCR (==5).
---
static int override_return_code(int dimm, unsigned int func, int rc)
{
if ((1 << func) & dimm_fail_cmd_flags[dimm]) {dimm_fail_cmd_flags[] definition:
static unsigned long dimm_fail_cmd_flags[NUM_DCR];
---
'dimm' is the return value of get_dimm(), and get_dimm() returns the index of handle[] array. The handle[] has 7 index, and the index #0 to #4 is for nfit_test.0 and #5, #6 is for nfit_test.1. NUM_DCR is only for nfit_test.0. Let's add for nfit_test.1.
KASAN report:
==================================================================
BUG: KASAN: global-out-of-bounds in nfit_test_ctl+0x47bb/0x55b0 [nfit_test] Read of size 8 at addr ffffffffc10cbbe8 by task kworker/u41:0/8 ... Call Trace: dump_stack+0xea/0x1b0 ? dump_stack_print_info.cold.0+0x1b/0x1b ? kmsg_dump_rewind_nolock+0xd9/0xd9 print_address_description+0x65/0x22e ? nfit_test_ctl+0x47bb/0x55b0 [nfit_test] kasan_report.cold.6+0x92/0x1a6 nfit_test_ctl+0x47bb/0x55b0 [nfit_test] ... The buggy address belongs to the variable: dimm_fail_cmd_flags+0x28/0xffffffffffffa440 [nfit_test] ==================================================================
--- tools/testing/nvdimm/test/nfit.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)
tools/testing/nvdimm/test/nfit.c +6 -4
@@ -104,6 +104,8 @@enum {NUM_PM =3 ,NUM_DCR =5 , +NUM_DCR_ALL =NUM_DCR +2 ,/* 5 DCRs for test.0, 2 DCRs for test.1 */ +NUM_HINTS =8 ,NUM_BDW =NUM_DCR ,NUM_SPA =NUM_PM +NUM_DCR +NUM_BDW ,
@@ -140,8 +142,8 @@ static u32 handle[] = { [6] = NFIT_DIMM_HANDLE(1, 0, 0, 0, 1), }; -static unsigned long dimm_fail_cmd_flags[NUM_DCR ]; -static int dimm_fail_cmd_code[NUM_DCR ]; +static unsigned long dimm_fail_cmd_flags[NUM_DCR_ALL ]; +static int dimm_fail_cmd_code[NUM_DCR_ALL ];static const struct nd_intel_smart smart_def = { .flags =ND_INTEL_SMART_HEALTH_VALID
@@ -205,7 +207,7 @@ struct nfit_test {unsigned long deadline;spinlock_t lock; } ars_state; -struct device *dimm_dev[NUM_DCR ]; +struct device *dimm_dev[NUM_DCR_ALL ];struct nd_intel_smart *smart;struct nd_intel_smart_threshold *smart_threshold;struct badrange badrange;
@@ -2680,7 +2682,7 @@ static int nfit_test_probe(struct platform_device *pdev)u32 nfit_handle =__to_nfit_memdev (nfit_mem)->device_handle;int i; -for (i =0 ; i <NUM_DCR ; i++) +for (i =0 ; i <NUM_DCR_ALL ; i++)if (nfit_handle == handle[i])dev_set_drvdata (nfit_test->dimm_dev[i], nfit_mem);
Signature · 6 lines
2.18.0 _______________________________________________ Linux-nvdimm mailing list [email protected] https://lists.01.org/mailman/listinfo/linux-nvdimm
Quoted 24 lines · Masayoshi Mizuma
From: Masayoshi Mizuma <[email protected]>
KASAN reports following global out of bounds access while
nfit_testis being loaded. The out of bound access happens the following reference to dimm_fail_cmd_flags[dimm]. 'dimm' is over than the index value,NUM_DCR(==5).---
static intoverride_return_code(int dimm, unsigned int func, int rc){ if ((1 << func) & dimm_fail_cmd_flags[dimm]) {dimm_fail_cmd_flags[] definition:
static unsigned long dimm_fail_cmd_flags[NUM_DCR];
---'dimm' is the return value of
get_dimm(), andget_dimm()returns the index of handle[] array. The handle[] has 7 index, and the index #0 to #4 is for nfit_test.0 and #5, #6 is for nfit_test.1.NUM_DCRis only for nfit_test.0. Let's add for nfit_test.1.KASAN report:
==================================================================
Quoted 15 lines
BUG: KASAN: global-out-of-bounds in nfit_test_ctl+0x47bb/0x55b0 [nfit_test] Read of size 8 at addr ffffffffc10cbbe8 by task kworker/u41:0/8 ... Call Trace: dump_stack+0xea/0x1b0 ? dump_stack_print_info.cold.0+0x1b/0x1b ? kmsg_dump_rewind_nolock+0xd9/0xd9 print_address_description+0x65/0x22e ? nfit_test_ctl+0x47bb/0x55b0 [nfit_test] kasan_report.cold.6+0x92/0x1a6 nfit_test_ctl+0x47bb/0x55b0 [nfit_test] ... The buggy address belongs to the variable: dimm_fail_cmd_flags+0x28/0xffffffffffffa440 [nfit_test] ==================================================================
Thanks, good catch. This mostly looks ok just one change request below...
Quoted 11 lines
Signed-off-by Masayoshi Mizuma <[email protected]>--- tools/testing/nvdimm/test/nfit.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)tools/testing/nvdimm/test/nfit.c +0
-0
tools/testing/nvdimm/test/nfit.c line 107
Quoted 4 lines
tools/testing/nvdimm/test/nfit.c +1
-0@@ -104,6 +104,8 @@enum {NUM_PM =3 ,NUM_DCR =5 , +NUM_DCR_ALL =NUM_DCR +2 ,/* 5 DCRs for test.0, 2 DCRs for test.1 */
Let's just use ARRAY_SIZE(handle) in place of introducing NUM_DCR_ALL.
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm
Quoted 9 lines · Masayoshi Mizuma
From: Masayoshi Mizuma <[email protected]>
KASAN reports following global out of bounds access while
nfit_testis being loaded. The out of bound access happens the following reference to dimm_fail_cmd_flags[dimm]. 'dimm' is over than the index value,NUM_DCR(==5).---
One more small comment, please be careful to not use "---" in the changelog because this causes "git am" to stop parsing the changelog right at this point. I lost your sign-off and debug information when test applying this.
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm
Quoted 9 lines · Dan Williams
Quoted 8 lines · Masayoshi Mizuma
From: Masayoshi Mizuma <[email protected]>
KASAN reports following global out of bounds access while
nfit_testis being loaded. The out of bound access happens the following reference to dimm_fail_cmd_flags[dimm]. 'dimm' is over than the index value,NUM_DCR(==5).
Quoted 6 lines
Quoted 1 line
---
One more small comment, please be careful to not use "---" in the changelog because this causes "git am" to stop parsing the changelog right at this point. I lost your sign-off and debug information when test applying this.
Sorry. Got it.
- Masa
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm