[PATCH] libnvdimm/labels: Fix divide error in nd_label_data_init()

[PATCH] libnvdimm/labels: Fix divide error in nd_label_data_init()

[Robert Richter]

If a CXL memory device returns a broken zero LSA size in its memory
device information (Identify Memory Device (Opcode 4000h), CXL
spec. 3.1, 8.2.9.9.1.1), a divide error occurs in the libnvdimm
driver:

 Oops: divide error: 0000 [#1] PREEMPT SMP NOPTI
 RIP: 0010:nd_label_data_init+0x10e/0x800 [libnvdimm]

Code and flow:

1) CXL Command 4000h returns LSA size = 0,
2) config_size is assigned to zero LSA size (CXL pmem driver):

drivers/cxl/pmem.c:             .config_size = mds->lsa_size,

3) max_xfer is set to zero (nvdimm driver):

drivers/nvdimm/label.c: max_xfer = min_t(size_t, ndd->nsarea.max_xfer, config_size);
drivers/nvdimm/label.c: if (read_size < max_xfer) {
drivers/nvdimm/label.c-         /* trim waste */

4) DIV_ROUND_UP() causes division by zero:

drivers/nvdimm/label.c:         max_xfer -= ((max_xfer - 1) - (config_size - 1) % max_xfer) /
drivers/nvdimm/label.c:                     DIV_ROUND_UP(config_size, max_xfer);
drivers/nvdimm/label.c-         /* make certain we read indexes in exactly 1 read */
drivers/nvdimm/label.c:         if (max_xfer < read_size)
drivers/nvdimm/label.c:                 max_xfer = read_size;
drivers/nvdimm/label.c- }

Fix this by checking the config size parameter by extending an
existing check.

Signed-off-by: Robert Richter <rrichter@amd.com>
---
 drivers/nvdimm/label.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/nvdimm/label.c b/drivers/nvdimm/label.c
index 082253a3a956..04f4a049599a 100644
--- a/drivers/nvdimm/label.c
+++ b/drivers/nvdimm/label.c
@@ -442,7 +442,8 @@ int nd_label_data_init(struct nvdimm_drvdata *ndd)
 	if (ndd->data)
 		return 0;
 
-	if (ndd->nsarea.status || ndd->nsarea.max_xfer == 0) {
+	if (ndd->nsarea.status || ndd->nsarea.max_xfer == 0 ||
+	    ndd->nsarea.config_size == 0) {
 		dev_dbg(ndd->dev, "failed to init config data area: (%u:%u)\n",
 			ndd->nsarea.max_xfer, ndd->nsarea.config_size);
 		return -ENXIO;
-- 
2.39.5

Re: [PATCH] libnvdimm/labels: Fix divide error in nd_label_data_init()

[Gupta, Pankaj]

On 3/19/2025 12:32 PM, Robert Richter wrote:
> If a CXL memory device returns a broken zero LSA size in its memory
> device information (Identify Memory Device (Opcode 4000h), CXL
> spec. 3.1, 8.2.9.9.1.1), a divide error occurs in the libnvdimm
> driver:
> 
>   Oops: divide error: 0000 [#1] PREEMPT SMP NOPTI
>   RIP: 0010:nd_label_data_init+0x10e/0x800 [libnvdimm]
> 
> Code and flow:
> 
> 1) CXL Command 4000h returns LSA size = 0,
> 2) config_size is assigned to zero LSA size (CXL pmem driver):
> 
> drivers/cxl/pmem.c:             .config_size = mds->lsa_size,
> 
> 3) max_xfer is set to zero (nvdimm driver):
> 
> drivers/nvdimm/label.c: max_xfer = min_t(size_t, ndd->nsarea.max_xfer, config_size);
> drivers/nvdimm/label.c: if (read_size < max_xfer) {
> drivers/nvdimm/label.c-         /* trim waste */
> 
> 4) DIV_ROUND_UP() causes division by zero:
> 
> drivers/nvdimm/label.c:         max_xfer -= ((max_xfer - 1) - (config_size - 1) % max_xfer) /
> drivers/nvdimm/label.c:                     DIV_ROUND_UP(config_size, max_xfer);
> drivers/nvdimm/label.c-         /* make certain we read indexes in exactly 1 read */
> drivers/nvdimm/label.c:         if (max_xfer < read_size)
> drivers/nvdimm/label.c:                 max_xfer = read_size;
> drivers/nvdimm/label.c- }
> 
> Fix this by checking the config size parameter by extending an
> existing check.
> 
> Signed-off-by: Robert Richter <rrichter@amd.com>

LGTM

Reviewed-by: Pankaj Gupta <pankaj.gupta@amd.com>

> ---
>   drivers/nvdimm/label.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/nvdimm/label.c b/drivers/nvdimm/label.c
> index 082253a3a956..04f4a049599a 100644
> --- a/drivers/nvdimm/label.c
> +++ b/drivers/nvdimm/label.c
> @@ -442,7 +442,8 @@ int nd_label_data_init(struct nvdimm_drvdata *ndd)
>   	if (ndd->data)
>   		return 0;
>   
> -	if (ndd->nsarea.status || ndd->nsarea.max_xfer == 0) {
> +	if (ndd->nsarea.status || ndd->nsarea.max_xfer == 0 ||
> +	    ndd->nsarea.config_size == 0) {
>   		dev_dbg(ndd->dev, "failed to init config data area: (%u:%u)\n",
>   			ndd->nsarea.max_xfer, ndd->nsarea.config_size);
>   		return -ENXIO;

Re: [PATCH] libnvdimm/labels: Fix divide error in nd_label_data_init()

[Ira Weiny]

Robert Richter wrote:
> If a CXL memory device returns a broken zero LSA size in its memory
> device information (Identify Memory Device (Opcode 4000h), CXL
> spec. 3.1, 8.2.9.9.1.1), a divide error occurs in the libnvdimm
> driver:
> 
>  Oops: divide error: 0000 [#1] PREEMPT SMP NOPTI
>  RIP: 0010:nd_label_data_init+0x10e/0x800 [libnvdimm]
> 
> Code and flow:
> 
> 1) CXL Command 4000h returns LSA size = 0,
> 2) config_size is assigned to zero LSA size (CXL pmem driver):
> 
> drivers/cxl/pmem.c:             .config_size = mds->lsa_size,
> 
> 3) max_xfer is set to zero (nvdimm driver):
> 
> drivers/nvdimm/label.c: max_xfer = min_t(size_t, ndd->nsarea.max_xfer, config_size);
> drivers/nvdimm/label.c: if (read_size < max_xfer) {
> drivers/nvdimm/label.c-         /* trim waste */
> 
> 4) DIV_ROUND_UP() causes division by zero:
> 
> drivers/nvdimm/label.c:         max_xfer -= ((max_xfer - 1) - (config_size - 1) % max_xfer) /
> drivers/nvdimm/label.c:                     DIV_ROUND_UP(config_size, max_xfer);

I think this is the wrong DIV_ROUND_UP which is failing because read_size is
never less than max_xfer is it?

I believe the failing DIV_ROUND_UP is after if statement here:

 489         /* Make our initial read size a multiple of max_xfer size */
 490         read_size = min(DIV_ROUND_UP(read_size, max_xfer) * max_xfer,
 491                         config_size);

Apparently nvdimm_get_config_data() was intended to check for this implicitly
but it is too late.

Anyway all this side tracked me a bit.

I assume this is a broken device which is in the real world?  The fix looks
fine.  But could you re-spin with a clean up of the commit message and I'll
queue it up.

Reviewed-by: Ira Weiny <ira.weiny@intel.com>

[snip]

Re: [PATCH] libnvdimm/labels: Fix divide error in nd_label_data_init()

[Robert Richter]

On 19.03.25 14:33:54, Ira Weiny wrote:
> Robert Richter wrote:
> > If a CXL memory device returns a broken zero LSA size in its memory
> > device information (Identify Memory Device (Opcode 4000h), CXL
> > spec. 3.1, 8.2.9.9.1.1), a divide error occurs in the libnvdimm
> > driver:
> > 
> >  Oops: divide error: 0000 [#1] PREEMPT SMP NOPTI
> >  RIP: 0010:nd_label_data_init+0x10e/0x800 [libnvdimm]
> > 
> > Code and flow:
> > 
> > 1) CXL Command 4000h returns LSA size = 0,
> > 2) config_size is assigned to zero LSA size (CXL pmem driver):
> > 
> > drivers/cxl/pmem.c:             .config_size = mds->lsa_size,
> > 
> > 3) max_xfer is set to zero (nvdimm driver):
> > 
> > drivers/nvdimm/label.c: max_xfer = min_t(size_t, ndd->nsarea.max_xfer, config_size);
> > drivers/nvdimm/label.c: if (read_size < max_xfer) {
> > drivers/nvdimm/label.c-         /* trim waste */
> > 
> > 4) DIV_ROUND_UP() causes division by zero:
> > 
> > drivers/nvdimm/label.c:         max_xfer -= ((max_xfer - 1) - (config_size - 1) % max_xfer) /
> > drivers/nvdimm/label.c:                     DIV_ROUND_UP(config_size, max_xfer);
> 
> I think this is the wrong DIV_ROUND_UP which is failing because read_size is
> never less than max_xfer is it?
> 
> I believe the failing DIV_ROUND_UP is after if statement here:
> 
>  489         /* Make our initial read size a multiple of max_xfer size */
>  490         read_size = min(DIV_ROUND_UP(read_size, max_xfer) * max_xfer,
>  491                         config_size);

Yes, it is this one.

> 
> Apparently nvdimm_get_config_data() was intended to check for this implicitly
> but it is too late.
> 
> Anyway all this side tracked me a bit.
> 
> I assume this is a broken device which is in the real world?  The fix looks
> fine.  But could you re-spin with a clean up of the commit message and I'll
> queue it up.

Yes, it was caused by a faulty device.

Sure, will update description and resend.

> 
> Reviewed-by: Ira Weiny <ira.weiny@intel.com>

Thanks for review,

-Robert