by Bjorn Helgaas

[permalink] [raw]

Subject: Re: [PATCH v4 15/15] dmaengine: dw-edma: Add pcim_iomap_table return checker

[+cc Krzysztof]

From reading the subject, I thought you were adding a function to
check the return values, i.e., a "checker." But you're really adding
"checks" :)

On Wed, Feb 03, 2021 at 10:58:06PM +0100, Gustavo Pimentel wrote:
> Detected by CoverityScan CID 16555 ("Dereference null return")
>
> Signed-off-by: Gustavo Pimentel <[email protected]>
> ---
> drivers/dma/dw-edma/dw-edma-pcie.c | 15 +++++++++++++++
> 1 file changed, 15 insertions(+)
>
> diff --git a/drivers/dma/dw-edma/dw-edma-pcie.c b/drivers/dma/dw-edma/dw-edma-pcie.c
> index 686b4ff..7445033 100644
> --- a/drivers/dma/dw-edma/dw-edma-pcie.c
> +++ b/drivers/dma/dw-edma/dw-edma-pcie.c
> @@ -238,6 +238,9 @@ static int dw_edma_pcie_probe(struct pci_dev *pdev,
> dw->rd_ch_cnt = vsec_data.rd_ch_cnt;
>
> dw->rg_region.vaddr = pcim_iomap_table(pdev)[vsec_data.rg.bar];
> + if (!dw->rg_region.vaddr)
> + return -ENOMEM;

This doesn't seem quite right. If pcim_iomap_table() fails, it
returns NULL. But then we assign "vaddr = NULL[vsec_data.rg.bar]"
which dereferences the NULL pointer even before your test.

This "pcim_iomap_table(dev)[n]" pattern is extremely common. There
are over 100 calls of pcim_iomap_table(), and

$ git grep "pcim_iomap_table(.*)\[.*\]" | wc -l

says about 75 of them are of this form, where we dereference the
result before testing it.

> dw->rg_region.vaddr += vsec_data.rg.off;
> dw->rg_region.paddr = pdev->resource[vsec_data.rg.bar].start;
> dw->rg_region.paddr += vsec_data.rg.off;
> @@ -250,12 +253,18 @@ static int dw_edma_pcie_probe(struct pci_dev *pdev,
> struct dw_edma_block *dt_block = &vsec_data.dt_wr[i];
>
> ll_region->vaddr = pcim_iomap_table(pdev)[ll_block->bar];
> + if (!ll_region->vaddr)
> + return -ENOMEM;
> +
> ll_region->vaddr += ll_block->off;
> ll_region->paddr = pdev->resource[ll_block->bar].start;
> ll_region->paddr += ll_block->off;
> ll_region->sz = ll_block->sz;
>
> dt_region->vaddr = pcim_iomap_table(pdev)[dt_block->bar];
> + if (!dt_region->vaddr)
> + return -ENOMEM;
> +
> dt_region->vaddr += dt_block->off;
> dt_region->paddr = pdev->resource[dt_block->bar].start;
> dt_region->paddr += dt_block->off;
> @@ -269,12 +278,18 @@ static int dw_edma_pcie_probe(struct pci_dev *pdev,
> struct dw_edma_block *dt_block = &vsec_data.dt_rd[i];
>
> ll_region->vaddr = pcim_iomap_table(pdev)[ll_block->bar];
> + if (!ll_region->vaddr)
> + return -ENOMEM;
> +
> ll_region->vaddr += ll_block->off;
> ll_region->paddr = pdev->resource[ll_block->bar].start;
> ll_region->paddr += ll_block->off;
> ll_region->sz = ll_block->sz;
>
> dt_region->vaddr = pcim_iomap_table(pdev)[dt_block->bar];
> + if (!dt_region->vaddr)
> + return -ENOMEM;
> +
> dt_region->vaddr += dt_block->off;
> dt_region->paddr = pdev->resource[dt_block->bar].start;
> dt_region->paddr += dt_block->off;
> --
> 2.7.4
>

2021-02-09 06:35:50

Hi Gustavo,

[...]
> > That's true, there are a lot of drivers that don't verify that pointer.
> > What do you suggest?
> > 1) To remove the verification so that is aligned with the other drivers
> > 2) Leave it as is. Or even to add this verification to the other drivers?
> >
> > Either way, I will add the pcim_iomap_table(pdev) before this
> > instruction.
> [...]
>
> A lot of the drivers consume the value from pcim_iomap_table() at
> a given BAR index directly as-is, some check if the pointer they got
> back is not NULL, a very few also check if the address at a given index
> is not NULL.
>
> Given that the memory allocation for the table can fail, we ought to
> check for a NULL pointer. It's a bit worrying that people decided to
> consume the value it returns directly without any verification.
>
> I only found two drivers that perform this additional verification of
> checking whether the address at a given index is valid, as per:
>
> https://lore.kernel.org/linux-pci/YCLFTjZQ2bCfGC+J@rocinante/
>
> Personally, I would opt for (2), and then like you suggested send
> a separate series to update other drivers so that they also include the
> this NULL pointer check.
>
> But let's wait for Bjorn's take on this, though.

As per Bjorn's reply:

https://lore.kernel.org/linux-pci/20210209185246.GA494880@bjorn-Precision-5520/

These extra checks I proposed would be definitely too much, especially
since almost everyone who uses pcim_iomap_table() also calls either
pcim_iomap_regions() or pcim_iomap_regions_request_all() before
accessing the table.

There probably is also an opportunity to simplify some of the other
drivers in the future, especially if do some API changes as per what
Bjorn suggested.

Sorry for taking your time, and thank you again!

Krzysztof