On Mon, Jul 8, 2024 at 6:52 PM Shani Peretz <shper...@nvidia.com> wrote:
>
> DPDK allows for two ways to specify PCI device numbers:
> a full version ("0000:08:00.0") and a short version ("08:00.0").
> The problem arises when the application uses one format (e.g., full)
> when running testpmd, but then tries to use the other format (e.g., short)
> in a subsequent command, leading to a failure.
>
> The cmp_dev_name func, which is responsible for comparing PCI device names,
> is not handling the inconsistent PCI number representations correctly.
> The suggested fix is to use the pci_parse function, which can parse
> the PCI device name and fill a struct rte_pci_addr with the standardized
> representation of the PCI number.
> By comparing the struct rte_pci_addr instances instead of the string
> representations, the application can ensure consistent handling of
> PCI device numbers, regardless of the format used.
>
> Fixes: a3ee360f4440 ("eal: add hotplug add/remove device")
> Cc: jblu...@infradead.org
>
> Signed-off-by: Shani Peretz <shper...@nvidia.com>
> Acked-by: Dariusz Sosnowski <dsosnow...@nvidia.com>
> Acked-by: Thomas Monjalon <tho...@monjalon.net>I find it strange that Thomas acked this patch (for example, the commit title prefix is wrong). I don't understand the issue. Please provide a reproducer. And ideally we need a unit test to track regressions on this topic. Thanks. -- David Marchand
