nvme bug?? -- clearing extra memory

nvme bug?? -- clearing extra memory

[Kallol Biswas]

Hi,
   I have come across the following lines in the 5.8.0 nvme driver code.

drivers/nvme/host/pci.c

iod->sg = mempool_alloc(dev->iod_mempool, GFP_ATOMIC);
if (!iod->sg)
       return BLK_STS_RESOURCE;
sg_init_table(iod->sg, blk_nr_phys_segments(req));

----------------------------------
void sg_init_table(struct scatterlist *sgl, unsigned int nents)
{
     memset(sgl, 0, sizeof(*sgl) * nents);
.........................................

Question: mempool_alloc will allocate a single sg entry, and then if
blk_nr_phys_segmets() is > 1, then we will clear more memory area than
intended.

Is my understanding correct?

-- 
------
Kallol Biswas
Phone: 408-718-8164 (c)
Phone: 408-725-7527 (o)
NucleoDyne Systems, Inc.

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: nvme bug?? -- clearing extra memory

[Keith Busch]

On Wed, Mar 31, 2021 at 12:39:42PM -0700, Kallol Biswas wrote:
> Hi,
>    I have come across the following lines in the 5.8.0 nvme driver code.
> 
> drivers/nvme/host/pci.c
> 
> iod->sg = mempool_alloc(dev->iod_mempool, GFP_ATOMIC);
> if (!iod->sg)
>        return BLK_STS_RESOURCE;
> sg_init_table(iod->sg, blk_nr_phys_segments(req));
> 
> ----------------------------------
> void sg_init_table(struct scatterlist *sgl, unsigned int nents)
> {
>      memset(sgl, 0, sizeof(*sgl) * nents);
> .........................................
> 
> Question: mempool_alloc will allocate a single sg entry, and then if
> blk_nr_phys_segmets() is > 1, then we will clear more memory area than
> intended.
> 
> Is my understanding correct?

The size from this mempool is set by nvme_pci_iod_alloc_size(), which is
larger than one sg entry.

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme