Thread (5 messages) 5 messages, 2 authors, 2012-07-23

Re: [PATCH] Prevent interface errors with Seagate FreeAgent GoFlex

From: Daniel J Blueman <hidden>
Date: 2012-05-04 02:44:54
Also in: lkml

On 4 May 2012 05:23, Martin Mokrejs [off-list ref] wrote:
Daniel J Blueman wrote:
quoted
When connected by the optional eSATAp cable, various interface errors
occur with my new external Seagate FreeAgent GoFlex (firmware 0110) until
the interface is forced to 1.5Gbps mode. Blacklist 3.0Gbps mode with it
to avoid the error messages, delays and dataloss potential.
[]
 maybe I am talking nonsense but recently there were some fixes to bus speed posted here at
linux-ide list from Lin Ming [off-list ref].

In 3.4-rc5 it is already fixed so that if you unplug and re-plug a SATA disk it can be detected
at 3.0Gbps instead of just 1.5 (because previous unplug caused errors). An additional patch
"Subject: [PATCH v2] libata: skip old error history when counting probe trials"
was just applied few hours ago some maybe you want to re-test? So is your blacklisting necessary?
This is an unrelated issue. The problems with the FreeAgent GoFlex
eSATAp optional cable are that it has buggy SATA2 support, and there
is a 100% correlation with errors regularly but infrequently (eg once
every few minutes) occurring at 3Gbps and no errors occurring at
1.5Gbps. The Seagate support forums tell the tale also (but largely on
Windows).
When inspecting my "lspci -vvvxx" outputs I see that lots of my chipset devices in my SandyBridge
laptop lifted up with the above mentioned patch from 2.5Gbps to 5Gbps
(was about in a thread "Re: Hotplug borked after suspend/resume in Linux-3.3 ?") which did not
original tackle the error counter issue which broke the re-plugging.
Another patch will have correctly restored the PCIe 2.0 link training
registers, so this isn't connected.
(I still do see problems with 3.4-rc5 on ExpressCard housed sata_sil24 so that unplug of a disk
and its re-plug results still only in 1.5Gbps instead of 3.0 Gbps, but if one waits a while
so that the SCSI device is removed, then a re-plug catches up at 3 Gbps. Probably have to apply
the patch to null the error counter. ;-))
You're observing the ~5 second delay on unplug before the disk is
considered removed, to prevent dropping disks with intermittent
link/power/controller reset issues, so here 1.5Gbps link state is held
if unplugged for less than ~5 seconds.

Thanks!
  Daniel
-- 
Daniel J Blueman
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help