netdev - Re: [PATCH 1/1] i40e: Avoid double IRQ free on error path in probe()

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

Message-ID: <50c21a769633c8efa07f49fc8b20fdfb544cf3c5.camel@intel.com>
Date:   Mon, 30 Aug 2021 20:52:41 +0000
From:   "Nguyen, Anthony L" <anthony.l.nguyen@...el.com>
To:     "pwaskiewicz@...ptrading.com" <pwaskiewicz@...ptrading.com>,
        "intel-wired-lan@...ts.osuosl.org" <intel-wired-lan@...ts.osuosl.org>
CC:     "pjwaskiewicz@...il.com" <pjwaskiewicz@...il.com>,
        "Loktionov, Aleksandr" <aleksandr.loktionov@...el.com>,
        "Fijalkowski, Maciej" <maciej.fijalkowski@...el.com>,
        "Dziedziuch, SylwesterX" <sylwesterx.dziedziuch@...el.com>,
        "davem@...emloft.net" <davem@...emloft.net>,
        "Brandeburg, Jesse" <jesse.brandeburg@...el.com>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: [PATCH 1/1] i40e: Avoid double IRQ free on error path in probe()

On Thu, 2021-08-26 at 17:19 -0500, PJ Waskiewicz wrote:
> This fixes an error path condition when probe() fails due to the
> default VSI not being available or online yet in the firmware. If
> that happens, the previous teardown path would clear the interrupt
> scheme, which also freed the IRQs with the OS. Then the error path
> for the switch setup (pre-VSI) would attempt to free the OS IRQs
> as well.

Hi PJ,

These comments are from the i40e team.

Yes in case we fail and go to err_vsis label in i40e_probe() we will
call i40e_reset_interrupt_capability twice but this is not a problem.
This is because pci_disable_msi/pci_disable_msix will be called only if
appropriate flags are set on PF and if this function is called ones it
will clear those flags. So even if we call
i40e_reset_interrupt_capability twice we will not disable msi vectors
twice.

The issue here is different however. It is failing in free_irq because
we are trying to free already free vector. This is because setup of
misc irq vectors in i40e_probe is done after i40e_setup_pf_switch. If
i40e_setup_pf_switch fails then we will jump to err_vsis and call
i40e_clear_interrupt_scheme which will try to free those misc irq
vectors which were not yet allocated. We should have the proper fix for
this ready soon.