linux-kernel - Re: [PATCH v3.5] dw_dmac: return proper residue value

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:	Fri, 25 Jan 2013 11:04:17 +0200
From:	Andy Shevchenko <andy.shevchenko@...il.com>
To:	Viresh Kumar <viresh.kumar@...aro.org>
Cc:	Vinod Koul <vinod.koul@...el.com>, linux-kernel@...r.kernel.org,
	spear-devel <spear-devel@...t.st.com>
Subject: Re: [PATCH v3.5] dw_dmac: return proper residue value

On Fri, Jan 25, 2013 at 10:56 AM, Andy Shevchenko
<andriy.shevchenko@...ux.intel.com> wrote:
> On Fri, 2013-01-25 at 09:43 +0530, Viresh Kumar wrote:
>> On Thu, Jan 24, 2013 at 6:37 PM, Andy Shevchenko
>> <andriy.shevchenko@...ux.intel.com> wrote:

>> > diff --git a/drivers/dma/dw_dmac.c b/drivers/dma/dw_dmac.c
>> > +static inline u32 dwc_get_residue(struct dw_dma_chan *dwc)
>> > +{
>> > +       unsigned long flags;
>> > +       u32 residue;
>> > +
>> > +       spin_lock_irqsave(&dwc->lock, flags);
>> > +
>> > +       residue = dwc->residue;
>>
>> you need to release the lock just here.
>>
>> > +       if (test_bit(DW_DMA_IS_SOFT_LLP, &dwc->flags) && residue)
>> > +               residue -= dwc_get_sent(dwc);
>>
>> why do you need lock while reading the control registers? It looks you didn't
>> get what i wanted to say earlier. We are using locks to protect some part for
>> shared data or critical sections. these are playing with dwc transfer
>> lists, etc..
>>
>> Probably we don't need a lock to read the control register as nobody can
>> guarantee that another access is not happening currently. As hardware is
>> changing this register continuously for the transfer.
>>
>> Let me know if i am missing something.

Okay, we have to have a protection here because get_sent reads two
registers consequentially. This means we could end up with scenario
with threads 1 and 2

1. read ctlhi
2. write ctlhi
2. write ctllo
1. read ctllo


-- 
With Best Regards,
Andy Shevchenko
--
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/