Hi,
And now for something completely different.
I started the libsata tests on one of the four machines I'll be doing
these tests on.
At the moment I'm testing on 'lilith', which is my multi-harddisk
server and playground.
At the moment I am hammering /dev/sda5, which is an ext4dev partition
with bonnie++.
sda is attached to my onboard ATA controller, driven by pata_via.
At the same moment I am doing a full resync of my MD RAID1 device with
two stripes.
Both stripes are on seperate 120GB disks, where one disk is attached
to a PCI controller, driven by pata_sil680 and one disk attached to an
onboard Promise ATA controller, driven by pata_pdc2027x.
I confirmed that all devices are driven by the new libsata drivers, it
can't be very much differentr since I disabled the entire "old" IDE
stack. :)
So, let's see how things go, so far so good.
For those who are interested, I attached the output of an lspci -vvv.
The file name has "stage1" included since I'll be adding my PCI cards
one by one to isolate problems to a single bus or card.
I hope my test input helps a bit.
Patrick
On Sat, 10 Feb 2007 17:22:28 +0100 Patrick Ale wrote:
> Hi,
>
> And now for something completely different.
>
> I started the libsata tests on one of the four machines I'll be doing
> these tests on.
Hi,
What are the libsata tests and where are they, please?
> At the moment I'm testing on 'lilith', which is my multi-harddisk
> server and playground.
>
> At the moment I am hammering /dev/sda5, which is an ext4dev partition
> with bonnie++.
> sda is attached to my onboard ATA controller, driven by pata_via.
So you are testing libata and ext4dev at the same time...
> At the same moment I am doing a full resync of my MD RAID1 device with
> two stripes.
> Both stripes are on seperate 120GB disks, where one disk is attached
> to a PCI controller, driven by pata_sil680 and one disk attached to an
> onboard Promise ATA controller, driven by pata_pdc2027x.
>
> I confirmed that all devices are driven by the new libsata drivers, it
> can't be very much differentr since I disabled the entire "old" IDE
> stack. :)
libsata == libata ?
> So, let's see how things go, so far so good.
>
> For those who are interested, I attached the output of an lspci -vvv.
> The file name has "stage1" included since I'll be adding my PCI cards
> one by one to isolate problems to a single bus or card.
>
> I hope my test input helps a bit.
---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
On 2/10/07, Randy Dunlap <[email protected]> wrote:
> Hi,
> What are the libsata tests and where are they, please?
No where, I just compile the drivers, see if my system boots up and
then I run some I/O intensive stuff, like bonnie++, emerge -e world at
the same time.
> So you are testing libata and ext4dev at the same time...
Not really, i dont focus on ext4dev, it just happens to be a partition
that's empty and has 40GB of space left for disk writes. I mostly use
XFS on my system.
>
> libsata == libata ?
I don't think there is a seperate libata, it's all put under libsata
if I understand the working correctly (Alan, don't sue me if I am
wrong).
Maybe it *IS* a nice idea to outline some standard tests that
developers want to have carried out or some specific tests to trace
bugs or test if a patch/fix does its work.. But that's not up to me, I
am not a coder, I can do some very basic stuff in C, that's where
programming ends for me, unless it's shell scripting.
Patrick
> > libsata == libata ?
> I don't think there is a seperate libata, it's all put under libsata
> if I understand the working correctly (Alan, don't sue me if I am
> wrong).
Right umm "Your honor the gentleman is human, this is not acceptable" ?
It is in drivers/ata the files are libata-* so it's normally known as
"libata", so libsata == libata.
> Maybe it *IS* a nice idea to outline some standard tests that
> developers want to have carried out or some specific tests to trace
> bugs or test if a patch/fix does its work.. But that's not up to me, I
> am not a coder, I can do some very basic stuff in C, that's where
> programming ends for me, unless it's shell scripting.
Beating the crap out of it using it for real work is one excellent form
of testing. We aren't yet at the point of only worrying about benchmark
tuning and corner cases.
Alan
On 2/10/07, Alan <[email protected]> wrote:
> Right umm "Your honor the gentleman is human, this is not acceptable" ?
Human and a looney who wishes Monty Python's Flying Circus still
existed so he could be a main character and having a great excuse to
act weird.
> Beating the crap out of it using it for real work is one excellent form
> of testing. We aren't yet at the point of only worrying about benchmark
> tuning and corner cases.
Exactly what I am doing :)
I know bonnie++ is mostly used for benchmarking, but it also is a
great tool to hammer down your hard-disks.
I also break RAID1 arrays volunteerly (spelling) and resync them to do
some stress tests on the disks, but I dont recommend others to do
these kind of tests, unless you dont worry about your data, or are a
fruitcake, like me.
Patrick
And again, for something completely different.
I am hammering my second machine now, a cool Siemens/Fujitsu laptop
(named elisabeth) with ICH7 ATA chipset and controller.
Currently the ATA controller is in "SATA" mode (ICH7 chipsets allow
you to run in native SATA mode or compatible mode, for the HP users,
compatible == legacy mode).
And for the interested, an lspci -vvv is attached of what goodies are
onboard and what exactly we're testing against.
Patrick
Hi,
On 10/02/07, Patrick Ale <[email protected]> wrote:
> And again, for something completely different.
>
> I am hammering my second machine now
You may want to use autotest
http://test.kernel.org/autotest
>
> Patrick
>
>
Regards,
Michal
--
Michal K. K. Piotrowski
LTG - Linux Testers Group (PL)
(http://www.stardust.webpages.pl/ltg/)
LTG - Linux Testers Group (EN)
(http://www.stardust.webpages.pl/linux_testers_group_en/)
So, my AMD Athlon XP2400+ with pata_pdc2027x, pata_sil680 and pata_via
survived four days of bonnie++, emerge -e world, dd if=//dev/random
of=/tmp/whatever, and most of all, four days of me :)
SO I guess it's safe to assume I didn't find any weird things by using
2.6.20 and the new libata drivers.
Now, since functionality by design is boring and my employer doesn't
torture me enough,, I' pulled the libsata-dev git and linux-2.6 git
and will see how that goes.
Tests on my Siemens laptop are still in progress, since i have to
reboot to windows every day (cooperate rules at work, network
authentication, yadayada).
I didnt have time yet to install Linux on my HP Pavilion dv8373 and on
my Dual P2 400Mhz with Intel 440BX motherboard, but I hope to do this
soon.
Have a joyfull day and i hope you have better weather than we have
here in Amsterdam.
Patrick Ale
Not even the best coder can -Dsense -Dsane me
Patrick Ale wrote:
> So, my AMD Athlon XP2400+ with pata_pdc2027x, pata_sil680 and pata_via
> Now, since functionality by design is boring and my employer doesn't
> torture me enough,, I' pulled the libsata-dev git and linux-2.6 git
> and will see how that goes.
It's libata. I mean, you yourself are testing several PATA drivers.
> Tests on my Siemens laptop are still in progress, since i have to
> reboot to windows every day (cooperate rules at work, network
> authentication, yadayada).
>
> I didnt have time yet to install Linux on my HP Pavilion dv8373 and on
> my Dual P2 400Mhz with Intel 440BX motherboard, but I hope to do this
> soon.
>
> Have a joyfull day and i hope you have better weather than we have
> here in Amsterdam.
I appreciate the testing. One interesting but tedious test would be
suspend tests (both to disk and ram). Most libata drivers using new EH
should be ready for them. Oh, well, there's should and there's reality.
Thanks a lot.
--
tejun
On 2/15/07, Tejun Heo <[email protected]> wrote:
> I appreciate the testing. One interesting but tedious test would be
> suspend tests (both to disk and ram). Most libata drivers using new EH
> should be ready for them. Oh, well, there's should and there's reality.
I will do these tests when I am at home, no problem :)
On 2/15/07, Tejun Heo <[email protected]> wrote:
Yeah and about the libsata/libata thingy, it is libata of course :)
libsata I will test later on, so be prepared allready for more ranting for me.
I even bought a cool sata controller with sata disk so I can mess around :D
Patrick
On 2/15/07, Patrick Ale <[email protected]> wrote:
> On 2/15/07, Tejun Heo <[email protected]> wrote:
>
> Yeah and about the libsata/libata thingy, it is libata of course :)
> libsata I will test later on, so be prepared allready for more ranting for me.
It is all libata! Whether pata or sata, it's still libata. The way the
distinction is made is by saying libata pata. So, there is _no_
libsata!
root@section 14:34:37 /sources/linux-2.6.20/
$ grep libsata * -r
root@section 14:37:49 /sources/linux-2.6.20/
All clear now?
On 2/15/07, Ioan Ionita <[email protected]> wrote:
>
> All clear now?
Yes sir, sorry.