Date:   Thu, 29 Jun 2017 17:16:41 +0200
From:   Greg KH <gregkh@...uxfoundation.org>
To:     "Luis R. Rodriguez" <mcgrof@...nel.org>
Cc:     jakub.kicinski@...ronome.com, nbroeking@...com,
        ming.lei@...hat.com, mfuzzey@...keon.com, ebiederm@...ssion.com,
        dmitry.torokhov@...il.com, wagi@...om.org, dwmw2@...radead.org,
        jewalt@...innovations.com, rafal@...ecki.pl,
        arend.vanspriel@...adcom.com, rjw@...ysocki.net,
        yi1.li@...ux.intel.com, atull@...nel.org, moritz.fischer@...us.com,
        pmladek@...e.com, johannes.berg@...el.com,
        emmanuel.grumbach@...el.com, luciano.coelho@...el.com,
        kvalo@...eaurora.org, luto@...nel.org,
        torvalds@...ux-foundation.org, keescook@...omium.org,
        takahiro.akashi@...aro.org, dhowells@...hat.com, pjones@...hat.com,
        hdegoede@...hat.com, alan@...ux.intel.com, tytso@....edu,
        paul.gortmaker@...driver.com, mtosatti@...hat.com,
        mawilcox@...rosoft.com, linux-kernel@...r.kernel.org,
        "[4.10+]" <stable@...r.kernel.org>
Subject: Re: [PATCH v2] firmware: fix batched requests - wake all waiters

On Mon, Jun 26, 2017 at 02:23:12PM -0700, Luis R. Rodriguez wrote:
> From: Jakub Kicinski <jakub.kicinski@...ronome.com>
> 
> The firmware cache mechanism serves two purposes, the secondary purpose is
> not well documented nor understood. This fixes a regression with the secondary
> purpose of the firmware cache mechanism: batched requests.
> 
> The firmware cache is used for:
> 
> 1) Addressing races with file lookups during the suspend/resume cycle
>    by keeping firmware in memory during the cycle
> 
> 2) Batched requests for the same file rely only on work from the first file
>    lookup, which keeps the firmware in memory until the last release_firmware()
>    is called
> 
> Batched requests *only* take effect if secondary requests come in prior to the
> first user calling release_firmware(). The devres name used for the internal
> firmware cache is used as a hint other pending requests are ongoing, the
> firmware buffer data is kept in memory until the last user of the buffer
> calls release_firmware(), therefore serializing requests and delaying the
> release until all requests are done.
> 
> Batched requests wait for a wakup or signal (we only accept SIGKILL now) so we
> can rely on the first file fetch to write to the pending secondary requests.
> Commit 5b029624948d ("firmware: do not use fw_lock for fw_state protection")
> ported the firmware API to use swait, and in doing so failed to convert
> complete_all() to swake_up_all() -- it used swake_up(), loosing the ability
> for *some* batched requests to take effect.
> 
> Without this fix it has been reported plugging in two Intel 6260 Wifi cards
> on a system will end up enumerating the two devices only 50% of the time
> [0]. The ported swake_up() should have actually two devices, however,
> *if more than two cards are used* the swake_up() would not suffice. This
> change is only part of the required fixes for batched requests. Subsequent
> fixes will follow.
> 
> This particular change should fix the cases where more than three requests
> with the same firmware name is used, otherwise batched requests will wait for
> MAX_SCHEDULE_TIMEOUT and just timeout eventually.
> 
> [0] https://bugzilla.kernel.org/show_bug.cgi?id=195477
> 
> Fixes: 5b029624948d ("firmware: do not use fw_lock for fw_state protection")
> CC: <stable@...r.kernel.org>    [4.10+]
> Cc: Ming Lei <ming.lei@...hat.com>
> Signed-off-by: Jakub Kicinski <jakub.kicinski@...ronome.com>
> [mcgrof: expanded on impact on commit log]
> Signed-off-by: Luis R. Rodriguez <mcgrof@...nel.org>
> ---
> 
> Greg, I think it would make sense to queue this in after the signal stable
> fixes [1].

As I just dropped them, can you redo this based on Linus's tree now?

thanks,

greg k-h

Powered by blists - more mailing lists