On 2019-12-18 11:24 a.m., Bjorn Helgaas wrote:
> On Wed, Dec 18, 2019 at 12:00:04PM -0500, Qian Cai wrote:
>> The commit a7d06153eea2 ("PCI: Fix pci_add_dma_alias() bitmask size")
>> introduced a compilation warning and a potential infinite loop because
>> it is no longer possible to be self-terminated as u8 is always less than
>> 256,
>>
>> In file included from ./include/linux/kernel.h:12,
>> from ./include/asm-generic/bug.h:19,
>> from ./arch/x86/include/asm/bug.h:83,
>> from ./include/linux/bug.h:5,
>> from ./include/linux/jump_label.h:250,
>> from ./arch/x86/include/asm/string_64.h:6,
>> from ./arch/x86/include/asm/string.h:5,
>> from ./include/linux/string.h:20,
>> from ./include/linux/uuid.h:12,
>> from ./include/linux/mod_devicetable.h:13,
>> from ./include/linux/pci.h:27,
>> from drivers/pci/search.c:11:
>> drivers/pci/search.c: In function 'pci_for_each_dma_alias':
>> ./include/linux/bitops.h:30:13: warning: comparison is always true due
>> to limited range of data type [-Wtype-limits]
>> (bit) < (size); \
>> ^
>> drivers/pci/search.c:46:3: note: in expansion of macro 'for_each_set_bit'
>> for_each_set_bit(devfn, pdev->dma_alias_mask, MAX_NR_DEVFNS) {
>>
>> Fixed it by using u16 for "devfn" in this occasion.
>
> Ugh, that is pretty subtle! Would you mind if we used "unsigned int"
> instead of "u16"? "u16" makes it look like just a mistake -- somebody
> is likely to come along and say devfn only needs "u8" and try to
> change it back. The same might happen with "unsigned int", but at
> least it doesn't look like it was chosen specifically to fit a devfn.
I agree: you're right that we really don't need an int that's
specifically sized here. There's really shouldn't be any gain in using a
smaller integer, in theory "int" should be the most "efficient" anyway,
just using a full register in hardware.
Logan
> Provisional patch below.
>
>> Fixes: a7d06153eea2 ("PCI: Fix pci_add_dma_alias() bitmask size")
>> Signed-off-by: Qian Cai <cai@xxxxxx>
>> ---
>> drivers/pci/search.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/pci/search.c b/drivers/pci/search.c
>> index 9e4dfae47252..42bc44d0e681 100644
>> --- a/drivers/pci/search.c
>> +++ b/drivers/pci/search.c
>> @@ -41,7 +41,7 @@ int pci_for_each_dma_alias(struct pci_dev *pdev,
>> * DMA, iterate over that too.
>> */
>> if (unlikely(pdev->dma_alias_mask)) {
>> - u8 devfn;
>> + u16 devfn;
>>
>> for_each_set_bit(devfn, pdev->dma_alias_mask, MAX_NR_DEVFNS) {
>> ret = fn(pdev, PCI_DEVID(pdev->bus->number, devfn),
>> --
>> 2.21.0 (Apple Git-122.2)
>
> commit f8bf2aeb651b ("PCI: Fix pci_add_dma_alias() bitmask size")
> Author: James Sewart <jamessewart@xxxxxxxxxx>
> Date: Tue Dec 10 15:51:33 2019 -0600
>
> PCI: Fix pci_add_dma_alias() bitmask size
>
> The number of possible devfns is 256, but pci_add_dma_alias() allocated a
> bitmap of size 255. Fix this off-by-one error.
>
> This fixes commits 338c3149a221 ("PCI: Add support for multiple DMA
> aliases") and c6635792737b ("PCI: Allocate dma_alias_mask with
> bitmap_zalloc()"), but I doubt it was possible to see a problem because
> it takes 4 64-bit longs (or 8 32-bit longs) to hold 255 bits, and
> bitmap_zalloc() doesn't save the 255-bit size anywhere.
>
> [bhelgaas: commit log, move #define to drivers/pci/pci.h, include loop
> limit fix from Qian Cai:
> https://lore.kernel.org/r/20191218170004.5297-1-cai@xxxxxx]
> Signed-off-by: James Sewart <jamessewart@xxxxxxxxxx>
> Signed-off-by: Bjorn Helgaas <bhelgaas@xxxxxxxxxx>
> Reviewed-by: Logan Gunthorpe <logang@xxxxxxxxxxxx>
>
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index e87196cc1a7f..7b5fa2eabe09 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -6017,7 +6017,7 @@ EXPORT_SYMBOL_GPL(pci_pr3_present);
> void pci_add_dma_alias(struct pci_dev *dev, u8 devfn)
> {
> if (!dev->dma_alias_mask)
> - dev->dma_alias_mask = bitmap_zalloc(U8_MAX, GFP_KERNEL);
> + dev->dma_alias_mask = bitmap_zalloc(MAX_NR_DEVFNS, GFP_KERNEL);
> if (!dev->dma_alias_mask) {
> pci_warn(dev, "Unable to allocate DMA alias mask\n");
> return;
> diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
> index a0a53bd05a0b..6394e7746fb5 100644
> --- a/drivers/pci/pci.h
> +++ b/drivers/pci/pci.h
> @@ -4,6 +4,9 @@
>
> #include <linux/pci.h>
>
> +/* Number of possible devfns: 0.0 to 1f.7 inclusive */
> +#define MAX_NR_DEVFNS 256
> +
> #define PCI_FIND_CAP_TTL 48
>
> #define PCI_VSEC_ID_INTEL_TBT 0x1234 /* Thunderbolt */
> diff --git a/drivers/pci/search.c b/drivers/pci/search.c
> index bade14002fd8..e4dbdef5aef0 100644
> --- a/drivers/pci/search.c
> +++ b/drivers/pci/search.c
> @@ -41,9 +41,9 @@ int pci_for_each_dma_alias(struct pci_dev *pdev,
> * DMA, iterate over that too.
> */
> if (unlikely(pdev->dma_alias_mask)) {
> - u8 devfn;
> + unsigned int devfn;
>
> - for_each_set_bit(devfn, pdev->dma_alias_mask, U8_MAX) {
> + for_each_set_bit(devfn, pdev->dma_alias_mask, MAX_NR_DEVFNS) {
> ret = fn(pdev, PCI_DEVID(pdev->bus->number, devfn),
> data);
> if (ret)
>
