linux-kernel - Re: PATCH: Leadtek DTVb1800H require longer delay for tuner reset

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] [day] [month] [year] [list]

Message-ID: <4D870FEF.6040009@suse.cz>
Date:	Mon, 21 Mar 2011 09:44:31 +0100
From:	Jiri Slaby <jslaby@...e.cz>
To:	Mirek Slugeň <thunder.m@...il.cz>
CC:	linux-dvb@...uxtv.org, linux-kernel@...r.kernel.org,
	Jiri Slaby <jirislaby@...il.com>
Subject: Re: PATCH: Leadtek DTVb1800H require longer delay for tuner reset

On 03/20/2011 11:05 PM, Mirek Slugeň wrote:
> Hi,
> 
>     I am using alot of tuners Leadtek DTV1800H under Linux, minimum 5
> devices per one PC with kernel 2.6.37.4. I discovered bug which I can se
> very often but not always. When device tuner is reseted there is not
> enough time to do all needed stuff and sometimes (randomly) kernel show
> this message:
> 
> [  821.369647] cx88[0]/1: IRQ loop detected, disabling interrupts
> 
> Patch is again very simple, we need to adjust time in reset function
> after cx_clear and cx_set calls from 50 to 75 us. I tested this patch on
> 4 PC with similar problems and all are without this issue.
> 
> M. Slugen
> 
> 
> kernel_cx88_reset.diff
> 
> 
> diff -Naur linux-2.6.37.4.old//drivers/media/video/cx88/cx88-cards.c linux-2.6.37.4/drivers/media/video/cx88/cx88-cards.c
> --- linux-2.6.37.4.old//drivers/media/video/cx88/cx88-cards.c	2011-03-20 21:30:52.000000000 +0100
> +++ linux-2.6.37.4/drivers/media/video/cx88/cx88-cards.c	2011-03-20 22:25:54.710228001 +0100
> @@ -3019,9 +3019,9 @@
>  		cx_set(MO_GP1_IO, 0x1010);
>  		mdelay(50);
>  		cx_clear(MO_GP1_IO, 0x10);
> -		mdelay(50);
> +		mdelay(75);
>  		cx_set(MO_GP1_IO, 0x10);
> -		mdelay(50);
> +		mdelay(75);

Wow, 75 ms of busy-waiting. Since others use msleep, all these should be
converted to msleep too. Or is this an atomic context, so that the
others are broken (e.g. cx88_dvico_xc2028_callback)?

Also for instance I don't understand this code:
                cx_write(MO_GP2_IO, 0xcf7);
                mdelay(50);
                cx_write(MO_GP2_IO, 0xef5);
                mdelay(50);
                cx_write(MO_GP2_IO, 0xcf7);
                msleep(10);

Delays of 50 ms and sleep of 10 ms? Given [um]delay spin _at least_ the
time specified, what was the intention of this?

regards,
-- 
js
suse labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/