lkml.org 
[lkml]   [2023]   [Feb]   [15]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    From
    SubjectRE: [EXT] Re: [PATCH v0] qed/qed_dev: guard against a possible division by zero
    Date
    > -----Original Message-----
    > From: Daniil Tatianin <d-tatianin@yandex-team.ru>
    > Sent: Tuesday, February 14, 2023 12:53 PM
    > To: Simon Horman <simon.horman@corigine.com>
    > Cc: Ariel Elior <aelior@marvell.com>; Manish Chopra
    > <manishc@marvell.com>; David S. Miller <davem@davemloft.net>; Eric
    > Dumazet <edumazet@google.com>; Jakub Kicinski <kuba@kernel.org>; Paolo
    > Abeni <pabeni@redhat.com>; Yuval Mintz <Yuval.Mintz@qlogic.com>;
    > netdev@vger.kernel.org; linux-kernel@vger.kernel.org
    > Subject: [EXT] Re: [PATCH v0] qed/qed_dev: guard against a possible division
    > by zero
    >
    > External Email
    >
    > ----------------------------------------------------------------------
    >
    >
    > On 2/9/23 2:13 PM, Simon Horman wrote:
    > > On Thu, Feb 09, 2023 at 01:38:13PM +0300, Daniil Tatianin wrote:
    > >> Previously we would divide total_left_rate by zero if num_vports
    > >> happened to be 1 because non_requested_count is calculated as
    > >> num_vports - req_count. Guard against this by explicitly checking for
    > >> zero when doing the division.
    > >>
    > >> Found by Linux Verification Center (linuxtesting.org) with the SVACE
    > >> static analysis tool.
    > >>
    > >> Fixes: bcd197c81f63 ("qed: Add vport WFQ configuration APIs")
    > >> Signed-off-by: Daniil Tatianin <d-tatianin@yandex-team.ru>
    > >> ---
    > >> drivers/net/ethernet/qlogic/qed/qed_dev.c | 2 +-
    > >> 1 file changed, 1 insertion(+), 1 deletion(-)
    > >>
    > >> diff --git a/drivers/net/ethernet/qlogic/qed/qed_dev.c
    > >> b/drivers/net/ethernet/qlogic/qed/qed_dev.c
    > >> index d61cd32ec3b6..90927f68c459 100644
    > >> --- a/drivers/net/ethernet/qlogic/qed/qed_dev.c
    > >> +++ b/drivers/net/ethernet/qlogic/qed/qed_dev.c
    > >> @@ -5123,7 +5123,7 @@ static int qed_init_wfq_param(struct qed_hwfn
    > >> *p_hwfn,
    > >>
    > >> total_left_rate = min_pf_rate - total_req_min_rate;
    > >>
    > >> - left_rate_per_vp = total_left_rate / non_requested_count;
    > >> + left_rate_per_vp = total_left_rate / (non_requested_count ?: 1);
    > >
    > > I don't know if num_vports can be 1.
    > > But if it is then I agree that the above will be a divide by zero.
    > >
    > > I do, however, wonder if it would be better to either:
    > >
    > > * Treat this case as invalid and return with -EINVAL if num_vports is
    > > 1; or
    > I think that's a good idea considering num_vports == 1 is indeed an invalid
    > value.
    > I'd like to hear a maintainer's opinion on this.

    Practically, this flow will only hit with presence of SR-IOV VFs. In that case it's
    always expected to have num_vports > 1.

    > > * Skip both the calculation immediately above and the code
    > > in the if condition below, which is the only place where
    > > the calculated value is used, if num_vports is 1.
    > > I don't think the if clause makes much sense if num_vports is
    > > one.left_rate_per_vp is also used below the if clause, it is assigned
    > > to
    > .min_speed in a for loop. Looking at that code division by 1 seems to make
    > sense to me in this case.
    > >
    > >> if (left_rate_per_vp < min_pf_rate / QED_WFQ_UNIT) {
    > >> DP_VERBOSE(p_hwfn, NETIF_MSG_LINK,
    > >> "Non WFQ configured vports rate [%d Mbps] is less
    > than one
    > >> percent of configured PF min rate[%d Mbps]\n",
    > >> --
    > >> 2.25.1
    > >>
    \
     
     \ /
      Last update: 2023-03-27 00:24    [W:3.758 / U:0.400 seconds]
    ©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site