Bug 222250 (AbsintheSyringe) - Libata
Summary: Libata
Keywords:
Status: CLOSED NEXTRELEASE
Alias: AbsintheSyringe
Product: Fedora
Classification: Fedora
Component: kernel
Version: 6
Hardware: All
OS: Linux
medium
high
Target Milestone: ---
Assignee: Kernel Maintainer List
QA Contact: Brian Brock
URL:
Whiteboard: Not solved
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2007-01-11 02:28 UTC by Adnan Hodzic
Modified: 2007-11-30 22:11 UTC (History)
2 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2007-02-15 00:15:55 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
LSPCI output (20.91 KB, text/plain)
2007-01-11 03:26 UTC, Douglas Furlong
no flags Details
dmesg output (23.46 KB, text/plain)
2007-01-11 03:27 UTC, Douglas Furlong
no flags Details
hdparm output (514 bytes, text/plain)
2007-01-11 03:28 UTC, Douglas Furlong
no flags Details
lspci -vv, hdaparm /dev/sad, uname -r (14.18 KB, application/octet-stream)
2007-01-11 03:32 UTC, Adnan Hodzic
no flags Details
lspci, lspci -vv, hdaparm /dev/sad, uname -r (14.18 KB, application/octet-stream)
2007-01-11 03:34 UTC, Adnan Hodzic
no flags Details

Description Adnan Hodzic 2007-01-11 02:28:57 UTC
Description of problem:
When I'm copying from my dvd/rw to my SATA disc, whole system incredibly slow,
besides from that transfer speed is ~1.x mbps. I think the most things is
explained in additional info.

Version-Release number of selected component (if applicable):
2.6.18-1.2869.fc6xen

How reproducible:


Steps to Reproduce:
1.
2.
3.
  
Actual results:
Complete system slow down, to a point where even moving your mouse/touchpad is
so slow that cursor twitches. Transfer speed goes to minimum, copying a DVD
takes from 15 to 25 min.

Expected results:
To not even feel any slow down, to have a 5/6.x mbps transfer rate.

Additional info:
Yesterday I did my first switch to Fedora, before that I was a Slackware user. I
had a same problem with Slackware too, which I would solve by recompiling my
kernel and adding append="libata.atapi_enabled=1" to my lilo. 

I tried doing the same thing with grub on Fedora, except I just added it as
libata.atapi_enabled=1, but nothing happends.

Please also note that with my own compiled kernel on Slackware everything worked
right after adding that line to lilo. Also any other distro? I've tried Mandriva
2007.0 Free, that also worked right out of box, without even editting lilo or
grub or kernel for that matter.

I've also did a full system update, kernel was also updated, but without results.

Comment 1 Douglas Furlong 2007-01-11 03:21:18 UTC
Hello all.

I think I may be suffering from the same problem as Adnan.

I recently purchased a Gigabyte GA-965P-S3 and attempted to install FC6 on the
system.

I found that when connecting the drives to the 2port Intel SATA device, though
the drives DO appear as sda, sdb, the installer would crash when it reaches the
point of formating the drives.

If I connect the two drives to the 4 port intel SATA chipset, I was able to
install, however the time it takes to format is excessively long.

When the drives are connected to the 4 port chipset, they appear as HD devices
as opposed to SD... devices.

Once the OS was installed, trying to use the environment is difficult due to the
performance, any hard drive access results in the mouse stuttering as you move
it across the screen.

I have attached the output from lspci -vv, hdparm /dev/hda /dev/hda and dmesg.

Please let me know if any more info is required.

Comment 2 Douglas Furlong 2007-01-11 03:26:03 UTC
Created attachment 145308 [details]
LSPCI output

Generated with

lspci -vv > /tmp/lspci_output.txt

Comment 3 Douglas Furlong 2007-01-11 03:27:20 UTC
Created attachment 145309 [details]
dmesg output

generated with

demsg > /tmp/dmesg_output.txt

Comment 4 Douglas Furlong 2007-01-11 03:28:43 UTC
Created attachment 145310 [details]
hdparm output

Generated with

hdparm /dev/hda /dev/hdc > /tmp/hdparm_output.txt

Comment 5 Adnan Hodzic 2007-01-11 03:32:59 UTC
Created attachment 145311 [details]
lspci -vv, hdaparm /dev/sad, uname -r

If any additional info needed please let me know.

Comment 6 Adnan Hodzic 2007-01-11 03:34:13 UTC
Created attachment 145312 [details]
lspci, lspci -vv, hdaparm /dev/sad, uname -r

If any additional info needed please let me know.

Comment 7 Adnan Hodzic 2007-01-11 15:06:39 UTC
Bug solved. 

Instead of libata.atapi_enabled=1 in my grub I used combined_mode=libata line. 

Also to my /etc/modprobe.conf I added line option libata atapi_enabled=1
I really think you should add this line by default, it cant do that much harm,
think about it.

Working even the better I expected, transfer speed is 8/9.x mbps. 

Full report on:
http://absinthesyringe.blogspot.com/2007/01/fedora-6-libata-problem_11.html

Comment 8 Douglas Furlong 2007-01-11 16:26:53 UTC
Unfortunately this solution does not solve my problem with hard drive access.

Comment 9 Douglas Furlong 2007-01-16 12:38:15 UTC
is there any more information required for this bug's resolution?

Comment 10 Douglas Furlong 2007-01-16 13:55:47 UTC
I have just tried with the latest kernel for FC6 Updates Testing

http://download.fedora.redhat.com/pub/fedora/linux/core/updates/testing/6/i386/kernel-2.6.19-1.2895.fc6.i686.rpm

Now, when I attempt to boot the system, I receive the following errors.

hda: maz request size: 512KiB
hda: lost interrupt
(repeated 4 times)
hda: Host Protected area detected.
current capacity is 625140335 sextors (320071 MB)
native capacity is 625142448 sectors (320072 MB)
hda: lost interrupt
hda: Host Protected area detected.
hda: 625142448 sectors (320072MB) w/ etc...

This continues and then moves on to hdc, and repeat's with the same errors.

it takes about 5 minutes to work all the way through the above.

The system appears to continue to then go through the boot process, repeating
the lost interupt for hdc (I don't seem to see any thing for hda), however after
another 10 minutes or so, I stopped the boot.

Comment 11 Douglas Furlong 2007-01-16 18:11:19 UTC
I have this resolved, to the extent that transfer speeds are now fairly normal.

Adding hda=noprobe hdc=noprobe to the kernel boot perameters resolved the issue.

Drives are now appearing as sda and sdb respectively, and transfer rates are
around 50MB.

I am not sure what you wish to do with this bug now though, I'm assuming some
thing still needs resolving.

Comment 12 Dave Jones 2007-02-15 00:15:55 UTC
adnan, libata.atapi_enabled=1 has been the default for some time now.

Douglas, there's been a sizable amount of libata changes in 2.6.20 that may help
your situation, and I'll be pushing out another FC6 update rebasing to this
hopefully early next week.  In the meantime, stick with the noprobe workaround
you have.

Reopen/file a new bug if it's still problematic after that update.



Note You need to log in before you can comment on or make changes to this bug.