linux-kernel - Re: [PATCH] usb: host: xhci: parameterize Renesas delay/retry

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: <5a057964ae1e26ed934035c183bf2967@posteo.net>
Date:   Mon, 19 Jun 2023 10:12:36 +0000
From:   Anne Macedo <retpolanne@...teo.net>
To:     Christian Lamparter <chunkeey@...il.com>
Cc:     Mathias Nyman <mathias.nyman@...el.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Yoshihiro Shimoda <yoshihiro.shimoda.uh@...esas.com>,
        Vinod Koul <vkoul@...nel.org>,
        Christian Lamparter <chunkeey@...glemail.com>,
        linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] usb: host: xhci: parameterize Renesas delay/retry

On 19.06.2023 10:19, Christian Lamparter wrote:
> On 6/19/23 00:46, Anne Macedo wrote:
>> Cards based on Renesas uPD720202 have their firmware downloaded during
>> boot by xhci-pci. At this step, the status of the firmware is read and
>> it takes a while for this read to happen (up to a few seconds). The
>> macros RENESAS_RETRY and RENESAS_DELAY are used to retry reading this
>> status byte from PCI a few times. If it can't read the status byte in
>> RENESAS_RETRY tries, it times out.
>> 
>> However, since this may vary from card to card, these retry and delay
>> values need to be tweaked. In order to avoid having to patch the code 
>> to
>> change these values, CONFIG_USB_XHCI_PCI_RENESAS_RETRY and
>> CONFIG_USB_XHCI_PCI_RENESAS_DELAY are introduced.
>> 
>> If applied, this patch helps to fix errors such as:
>> 
>> ROM Download Step 34 failed at position 136 bytes
>> Firmware Download Step 2 failed at position 8 bytes with (-110)
>> 
>> while loading xhci-pci when using these cards.
>> 
>> This error in particular has been noticed by this e-mail [1].
>> 
>> [1] https://lore.kernel.org/lkml/20190626070658.GP2962@vkoul-mobl/
> 
> Can you tell me on what hardware (is it something older, with maybe
> a Synopsys/Designware PCIe host controller?) do you experience these
> errors and what delay+retry values are you configuring in order to
> get your DUT up an running?

It's a PH61 Rev 1.2 board with the Renesas uPD720201 host controller.
I'm using 10 as the delay and 6000 as the retry.
> 
> From what I can tell, the quoted [1] link to Vinod's mail was just
> an update during development. This was v3 of the patch series back
> then (and it went on to v10 I think, so this wasn't an issue with
> what's in the kernel right now).
> 

I see. That was the only reference I found to the timeout error I was
seeing, so that's why I decided to tweak these retry+delay values.

> Note: If you are interested I still got the "uPD720201/uPD720202 User's
> Manual" back then from Renesas site. (Nowadays they want you to
> register or something.). This document was the base for the code and
> maybe there's something in there you can quote to extend the
> retries/delays.

Definitely interested! I did find a .pdf on Google though, I can check 
it.

> 
> (From what I vaguely remember. Most of the transfer was fast and
> no retries where necessary, but some register write took so long.
> Vinod  also posted hints about a newer firmware for the
> uPD720201/uPD720202. Have you tried that as well?)

I was using the upd72020x-fw AUR package on my Arch Linux build,
if that's any relevant. However, it's quite old and I don't really know
how it works. I missed these hints, sorry, could you point me to
them?
> 
> Regards,
> Christian