linux-kernel - Re: [PATCH 2/3] hwmon: (pmbus/ltc2978): Fix PMBus polling of MFR

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]

Date:   Tue, 28 Jan 2020 11:13:06 -0800
From:   Guenter Roeck <linux@...ck-us.net>
To:     Mike Jones <michael-a1.jones@...log.com>
Cc:     linux-hwmon@...r.kernel.org, devicetree@...r.kernel.org,
        linux-doc@...r.kernel.org, linux-kernel@...r.kernel.org,
        jdelvare@...e.com, robh+dt@...nel.org, corbet@....net
Subject: Re: [PATCH 2/3] hwmon: (pmbus/ltc2978): Fix PMBus polling of
 MFR_COMMON definitions.

On Tue, Jan 28, 2020 at 10:59:59AM -0700, Mike Jones wrote:
> Change 21537dc driver PMBus polling of MFR_COMMON from bits 5/4 to
> bits 6/5. This fixs a LTC297X family bug where polling always returns
> not busy even when the part is busy. This fixes a LTC388X and
> LTM467X bug where polling used PEND and NOT_IN_TRANS, and BUSY was
> not polled, which can lead to NACKing of commands. LTC388X and
> LTM467X modules now poll BUSY and PEND, increasing reliability by
> eliminating NACKing of commands.
> 
> Signed-off-by: Mike Jones <michael-a1.jones@...log.com>

Fixes: e04d1ce9bbb49 ("hwmon: (ltc2978) Add polling for chips requiring it")

> ---
>  drivers/hwmon/pmbus/ltc2978.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/hwmon/pmbus/ltc2978.c b/drivers/hwmon/pmbus/ltc2978.c
> index f01f488..a91ed01 100644
> --- a/drivers/hwmon/pmbus/ltc2978.c
> +++ b/drivers/hwmon/pmbus/ltc2978.c
> @@ -82,8 +82,8 @@ enum chips { ltc2974, ltc2975, ltc2977, ltc2978, ltc2980, ltc3880, ltc3882,
>  
>  #define LTC_POLL_TIMEOUT		100	/* in milli-seconds */
>  
> -#define LTC_NOT_BUSY			BIT(5)
> -#define LTC_NOT_PENDING			BIT(4)
> +#define LTC_NOT_BUSY			BIT(6)
> +#define LTC_NOT_PENDING			BIT(5)
>  

In ltc_wait_ready(), we have:

	/*
         * LTC3883 does not support LTC_NOT_PENDING, even though
         * the datasheet claims that it does.
         */
        mask = LTC_NOT_BUSY;
        if (data->id != ltc3883)
                mask |= LTC_NOT_PENDING;

The semantics of this code is now different: It means that on
LTC3883 only bit 6 is checked; previously, it was bit 5. I agree
that the above change makes sense, but it doesn't seem correct
to drop the check for bit 5 on LTC3883. Maybe remove the if() above
and always check for bit 5 and 6 ? Or should bit 4 be checked
on parts other than LTC3883 ?

#define LTC_NOT_TRANSITIONING		BIT(4)
...
        mask = LTC_NOT_BUSY | LTC_NOT_PENDING;
        if (data->id != ltc3883)
                mask |= LTC_NOT_TRANSITIONING;

Thanks,
Guenter