linux-kernel - Re: watchdog: sp5100_tco support for AMD V/R/E series

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: <9e270546-7962-932b-2e4c-3c833b7d4b30@siemens.com>
Date:   Mon, 7 Sep 2020 17:46:37 +0200
From:   Jan Kiszka <jan.kiszka@...mens.com>
To:     Guenter Roeck <linux@...ck-us.net>, linux-watchdog@...r.kernel.org
Cc:     Wim Van Sebroeck <wim@...ux-watchdog.org>,
        linux-kernel@...r.kernel.org,
        "Awan, Arsalan" <Arsalan_Awan@...tor.com>,
        "Hombourger, Cedric" <Cedric_Hombourger@...tor.com>,
        "Farnsworth, Wade" <wade_farnsworth@...tor.com>
Subject: Re: watchdog: sp5100_tco support for AMD V/R/E series

On 07.09.20 17:31, Guenter Roeck wrote:
> On 9/7/20 4:20 AM, Jan Kiszka wrote:
>> Hi all,
>>
>> Arsalan reported that the upstream driver for sp5100_tco does not work
>> for embedded Ryzen. Meanwhile, I was able to confirm that on an R1505G:
>>
>> [   11.607251] sp5100_tco: SP5100/SB800 TCO WatchDog Timer Driver
>> [   11.607337] sp5100-tco sp5100-tco: Using 0xfed80b00 for watchdog MMIO address
>> [   11.607344] sp5100-tco sp5100-tco: Watchdog hardware is disabled
>>
>> ..and fix it:
>>
>> diff --git a/drivers/watchdog/sp5100_tco.c b/drivers/watchdog/sp5100_tco.c
>> index 85e9664318c9..5482154fde42 100644
>> --- a/drivers/watchdog/sp5100_tco.c
>> +++ b/drivers/watchdog/sp5100_tco.c
>> @@ -193,7 +193,8 @@ static void tco_timer_enable(struct sp5100_tco *tco)
>>  		/* Set the Watchdog timer resolution to 1 sec and enable */
>>  		sp5100_tco_update_pm_reg8(EFCH_PM_DECODEEN3,
>>  					  ~EFCH_PM_WATCHDOG_DISABLE,
>> -					  EFCH_PM_DECODEEN_SECOND_RES);
>> +					  EFCH_PM_DECODEEN_SECOND_RES |
>> +					  EFCH_PM_DECODEEN_WDT_TMREN);
> 
> Confusing. The register in question is a 32-bit register, but only a byte
> is written into it. Bit 24-25 are supposed to be the resolution, bit 25-26
> set to 0 enable the watchdog. Bit 7 is supposed to enable MMIO decoding.
> This is from AMD Publication 52740. So something in the existing code
> is (or seems to be) wrong, but either case I don't see how setting bit 7
> (or 31 ?) would enable the watchdog hardware.
> 
> Hmm, I wrote that code. Guess I'll need to to spend some time figuring out
> what is going on.

The logic came from [1] which inspired [2] - that's where I pointed out
the large overlap with the existing upstream driver. I would love to see
all that consolidated.

BTW, the R1505G is family 0x17. Maybe something changed there, and that
bit 7 was just reserved/ignored so far. ENOSPECS

Jan

[1]
https://git.yoctoproject.org/cgit/cgit.cgi/meta-amd/commit/meta-amd-bsp/recipes-kernel/amd-wdt/files/amd_wdt.c?id=cd760c9f04d276382a0f5156dabdb766594ace56
[2]
https://github.com/siemens/efibootguard/commit/3a702aa96d193f26571ea4e70db29ef01a0d4d5f

-- 
Siemens AG, Corporate Technology, CT RDA IOT SES-DE
Corporate Competence Center Embedded Linux