netdev - Re: [PATCH net-next] liquidio: use meaningful names for IRQs

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CA+sq2Ce5HgtxZAFNsA+rAgf2-VXfJQkFDWXdQ-L8aVaVvqTofg@mail.gmail.com>
Date:   Mon, 3 Apr 2017 23:12:14 +0530
From:   Sunil Kovvuri <sunil.kovvuri@...il.com>
To:     Ricardo Farrington <Ricardo.Farrington@...ium.com>
Cc:     "Manlunas, Felix" <Felix.Manlunas@...ium.com>,
        "David S. Miller" <davem@...emloft.net>,
        Linux Netdev List <netdev@...r.kernel.org>,
        "Vatsavayi, Raghu" <Raghu.Vatsavayi@...ium.com>,
        "Chickles, Derek" <Derek.Chickles@...ium.com>,
        "Burla, Satananda" <Satananda.Burla@...ium.com>
Subject: Re: [PATCH net-next] liquidio: use meaningful names for IRQs

On Mon, Apr 3, 2017 at 8:33 PM, Ricardo Farrington
<Ricardo.Farrington@...ium.com> wrote:
> Hi Sunil - this implementation was verified with multiple adapters installed.  In this case, the <id> field serves to differentiate the adapter.
>
>>     LiquidIO<id>-<func>-<type>-<queue pair num>
>
> Rick
>

No, what I was referring to was, let's say there are two adapters, if
we do a unbind of driver and
then bind driver in reverse sequence, then the <id> of each adapter
will get exchanged. isn't it ?

If the above is true, the logical interface's name will still be the
same but the <id> is different,
so it's not easy to map a logical interface to it's registered IRQs,
hence doing affinity setting is still problematic.

But moreover it seems the driver is calling 'octeon_allocate_device()'
for each VF, so not sure how <id> field which
is derived from 'oct->octeon_id' helps in differentiating the adapter.
Is there only one VF per adapter ?

821 static int
822 liquidio_vf_probe(struct pci_dev *pdev,
823                   const struct pci_device_id *ent __attribute__((unused)))
824 {
825         struct octeon_device *oct_dev = NULL;
826
827         oct_dev = octeon_allocate_device(pdev->device,
828                                          sizeof(struct octeon_device_priv));
829

Thanks,
Sunil.