linux-kernel - RE: sata_mv port lockup on hotplug (kernel 2.6.38.2)

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: <C1438B59050E1B4C9482FF3266AD6BA32C772DA5DD@gretna.indigovision.com>
Date:	Fri, 15 Apr 2011 14:21:56 +0100
From:	Bruce Stenning <b.stenning@...igovision.com>
To:	Mark Lord <kernel@...savvy.com>
CC:	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"linux-ide@...r.kernel.org" <linux-ide@...r.kernel.org>,
	Justin Maggard <jmaggard10@...il.com>
Subject: RE: sata_mv port lockup on hotplug (kernel 2.6.38.2)

> > I also tried out my tracing in writelfl (without any other changes) and I was
> > able to lock up a port without any apparent unusual changes to the register
> > containing the irq masks.  Are there other routes to disabling the irqs?
>
> Mmm.. presumably you are NOT using MSI interrupts, right?

Hi Mark,

That is correct:

sata_mv 0000:00:04.0: Gen-II 32 slots 4 ports SCSI mode IRQ via INTx

I am building with CONFIG_ARCH_SUPPORTS_MSI=y, but also with CONFIG_PCI_MSI
not set.

With all this discussion of SATA link speed, I ought to say that we limit our
SATA links to 1.5Gbps with the following kernel parameter:

libata.force=1.5g

I had noticed occasional messages similar to the following earlier in the week:

ata4.00: limiting speed to UDMA/100:PIO4

I was surprised to see them, and thought they might be related to the hotplug
issues, but I was able to reproduce the lockups without these messages being
generated.  In some cases (but not all) both sata_down_spd_limit and
ata_down_xfermask_limit are called when backing off the link speed.

Mark, I was intrigued by your comment from the following message:

http://www.spinics.net/lists/linux-ide/msg36922.html
> If it's like their non-AHCI controllers (sata_mv), then the chipset/phy
> could be very particular about the sequence/timing used when changing speeds.

Is it possible that the driver is doing some of the work to change the link
speed (even though it has nowhere to go) and clobbering the link entirely?
I shall take another closer look at the source code and tracing around hotplug
w.r.t. link speed.

Cheers,

Bruce.

Bruce Stenning,
IndigoVision,
b <dot> stenning <at> indigovision <dot> com

Latest News at: http://www.indigovision.com/index.php/en/news.html