[opensuse-kernel] Why is the kernel doing this?!
I have 2 identical HDDs (1TB Seagates) and a DVDWriter in my system and all are connected using SATA 3 cables. I couldn't but help noticing last night that transferring data between a new external SATA 3 HDD I bought yesterday and one of my internal drives was at ~130 MB/s as agaist ~27 MB/s with the second drive. So I went looking in /var/log/messages. Sure enough the kernel had decided that the second HDD has a 40-wire cable and has therefore configured it to have UDMA 33! The devices are shown as: ata1 max UDMA 133 configured as 133 UDMA ata5 max UDMA 133 configured as 33 UDMA (because connected using 40-wire cable) ata3 max UDMA 100 configured as 100 UDMA (this is the DVD writer) This was also occurring on my other 32-bit system (different devices of course) and there I was able to overcome this nonsense by adding to the kernel boot parameters- libata.force=1:80c,2:80c but doing similar on the system (see signature line below) the kernel locks up when the system is booting. How the heck can the kernel get it so wrong re the cabling (I know that the SATA and ATA use the same controller) and treats the second HDD as a poor relative? Is the kernel designed to only always know about one single HDD? Does anyone know, please, what I need to add in the kernel boot parameters to get this second HDD running at full UDMA speed? BC -- Using openSUSE 12.2 x86_64 KDE 4.9.4 & kernel 3.6.8-1 on a system with- AMD FX 8-core 3.6/4.2GHz processor 16GB PC14900/1866MHz Quad Channel Corsair "Vengeance" RAM Gigabyte AMD3+ m/board; Gigabyte nVidia GTX550Ti 1GB DDR5 GPU -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On 06/12/12 11:10, Basil Chupin wrote: [pruned]
libata.force=1:80c,2:80c
but doing similar on the system (see signature line below) the kernel locks up when the system is booting.
My apologies.... I did do this but on my wife's computer (my "old" 32-bit system) on which I installed openSUSE 12.2 several days ago and also used grub rather than grub2. I couldn't have tried it on my 64-bit system because - stupid me! - I am using grub2 on it :-( . So my final plea re my hassle is: what do I need to do to get the UDMA correct when using grub2, please? BC -- Using openSUSE 12.2 x86_64 KDE 4.9.4 & kernel 3.6.8-1 on a system with- AMD FX 8-core 3.6/4.2GHz processor 16GB PC14900/1866MHz Quad Channel Corsair "Vengeance" RAM Gigabyte AMD3+ m/board; Gigabyte nVidia GTX550Ti 1GB DDR5 GPU -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On 2012-12-06 11:38 (GMT++1100) Basil Chupin composed:
So my final plea re my hassle is: what do I need to do to get the UDMA correct when using grub2, please?
Grub does not determine how the kernel configures ATA. Progress is unlikely until this registers. Kernel options are the same regardless which Grub is used to load the kernel. Did you try moving the SATA cables among the various motherboard ports? If you swap cables, does the problem stay with the HD, or move to the other? Have you used the diagnostic software provided by the HD manufacturer to ensure absence of defects? Are you sure the external carrier isn't where the problem lies? This latter is a problem I had: https://bugzilla.novell.com/show_bug.cgi?id=757426 Comments in the bug or its upstream may help you figure out which if any cmdline option will help. -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On 06/12/12 12:11, Felix Miata wrote:
On 2012-12-06 11:38 (GMT++1100) Basil Chupin composed:
So my final plea re my hassle is: what do I need to do to get the UDMA correct when using grub2, please?
Grub does not determine how the kernel configures ATA.
I know that! But it is what you put into the parameters for the kernel in grub/grub2 which determines what the damn kernel will do regarding the configuration of the devices' UDMAs.
Progress is unlikely until this registers.
Ce`?
Kernel options are the same regardless which Grub is used to load the kernel.
And so, which parameter I put into ????: 'libata.force=5:80c' or 'force=5:80c'? But the bottom line is why *should* I even think about doing this fiddle in this day age and day? Why does the kernel crap out about the 40-wire thingie when such cables haven't been used for.....how many years?
Did you try moving the SATA cables among the various motherboard ports?
No. But why should I? I expect the kernel to know what the heck it is doing. I am not running any abnormal system which was imported from some planet in a galaxy yet to be discovered - it is a stock standard mobo and cpu, and openSUSE 12.2 which hundreds of other people would have installed (but obviously they just haven't paid attention to how their system is performing :-( ).
If you swap cables, does the problem stay with the HD, or move to the other?
No, haven't done this. But again, while it is an option to try, why should I do this?
Have you used the diagnostic software provided by the HD manufacturer to ensure absence of defects?
This is sounding like the list of mandatory questions any call-centre is given to ask when a customer complains about a service :-( . The HDDs haven't collapsed within the normal 3 months after purchase so there is nothing wrong with them.
Are you sure the external carrier isn't where the problem lies?
What "external carrier"? Surely I could not have "external carrier" problems on 2, or even 3, computers?
This latter is a problem I had: https://bugzilla.novell.com/show_bug.cgi?id=757426 Comments in the bug or its upstream may help you figure out which if any cmdline option will help.
OK, I have had a quick look (and I mean this :-) ) of that 'bug' report but simply cannot fathom why anybody has to go thru this rubbish of reporting such a "bug" when it is simply bad coding in the kernel! BC -- Using openSUSE 12.2 x86_64 KDE 4.9.4 & kernel 3.6.8-1 on a system with- AMD FX 8-core 3.6/4.2GHz processor 16GB PC14900/1866MHz Quad Channel Corsair "Vengeance" RAM Gigabyte AMD3+ m/board; Gigabyte nVidia GTX550Ti 1GB DDR5 GPU -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Basil Chupin wrote:
[...] OK, I have had a quick look (and I mean this :-) ) of that 'bug' report but simply cannot fathom why anybody has to go thru this rubbish of reporting such a "bug" when it is simply bad coding in the kernel!
your patches to improve the "bad coding" in the kernel are certainly welcome... I must say, though, that I find your tone inappropriately aggressive and demanding. Complaining to the people that try to help you won't get you very far. Holger -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On 2012-12-01 01:32 (GMT+1100) Basil Chupin composed:
On 06/12/12 12:11, Felix Miata wrote:
Surely I could not have "external carrier" problems on 2, or even 3, computers?
This latter is a problem I had: https://bugzilla.novell.com/show_bug.cgi?id=757426 Comments in the bug or its upstream may help you figure out which if any cmdline option will help.
OK, I have had a quick look (and I mean this :-) ) of that 'bug' report but simply cannot fathom why anybody has to go thru this rubbish of reporting such a "bug" when it is simply bad coding in the kernel!
That bug contains a reference to an upstream bug. Combined they explain that a gizmo that fits between HD and cable can be defective in a way kernels cannot deal with except to slow down the HD I/O. What do you prefer such a defective device do, slow down the data rate, or corrupt the data transferred? -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
participants (3)
-
Basil Chupin
-
Felix Miata
-
Holger Arnold