2.4.20-pre2-ac4 IDE is slow

2.4.20-pre2-ac4 IDE is slow

[Rob Speer]

I think I remember someone else bringing up this same issue, in which
case I'm sorry to have to ask again, but I can't find the message.

I'm going from 2.4.19 to 2.4.20-pre2-ac4 and the hard drive is noticably
slower in the new version. (It doesn't use DMA in either version - I
wish it did in ac4, but that's a separate problem.)

What I seem to remember from the other message is that there's some
parameter that can be changed to bring the speed back up. Could someone
tell me what it is?


If it helps: output of hdparm /dev/hda

/dev/hda:
 multcount    =  0 (off)
 IO_support   =  0 (default 16-bit)
 unmaskirq    =  0 (off)
 using_dma    =  0 (off)
 keepsettings =  0 (off)
 readonly     =  0 (off)
 readahead    =  8 (on)
 geometry     = 77557/16/63, sectors = 78177792, start = 0

-- 
Rob Speer

Re: 2.4.20-pre2-ac4 IDE is slow

[Samuel Flory]

  Can you send the ide section of dmesg?  What are the results of hdparm
-t /dev/hda on both kernels?


On Thu, 2002-08-22 at 10:59, Rob Speer wrote:
> I think I remember someone else bringing up this same issue, in which
> case I'm sorry to have to ask again, but I can't find the message.
> 
> I'm going from 2.4.19 to 2.4.20-pre2-ac4 and the hard drive is noticably
> slower in the new version. (It doesn't use DMA in either version - I
> wish it did in ac4, but that's a separate problem.)
> 
> What I seem to remember from the other message is that there's some
> parameter that can be changed to bring the speed back up. Could someone
> tell me what it is?
> 
> 
> If it helps: output of hdparm /dev/hda
> 
> /dev/hda:
>  multcount    =  0 (off)
>  IO_support   =  0 (default 16-bit)
>  unmaskirq    =  0 (off)
>  using_dma    =  0 (off)
>  keepsettings =  0 (off)
>  readonly     =  0 (off)
>  readahead    =  8 (on)
>  geometry     = 77557/16/63, sectors = 78177792, start = 0
> 
> -- 
> Rob Speer
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: 2.4.20-pre2-ac4 IDE is slow

[Rob Speer]

As many people have asked in private e-mail, I'm providing some actual
numbers to justify that the IDE is slower.

I was also told to set the hdparm options -a 1 -b 1 -c 1 -d 1 -u 1. -d 1
doesn't work, as DMA doesn't work (perhaps that would solve all of this
if it did).

The results from 'hdparm -t /dev/hda':

On 2.4.20-pre1:
  Without hdparm options: 2.5 MB/sec
  With hdparm options: 4.17 MB/sec

On 2.4.20-pre2-ac4 (with a1.patch so it doesn't kernel panic on startup):
  Without hdparm options: ???
  With hdparm options: 877.21 kB/sec

The ??? is there because it actually *lies* in this situation. It says
it took 10.52 seconds, giving 6.08 MB/sec, but it actually takes as long
as it does with the hdparm options, if not longer.

The thing is, on -ac4, the entire system becomes unresponsive while
hdparm is going on. The clock in the corner of my screen stops counting.
With hdparm on, the clock jumps forward the appropriate amount at the
end. Without hdparm, the clock only jumps forward... 10 seconds.

Could the hard disk be preventing even the system clock from working?

As I was saying, I suppose this would all be wonderfully fast if DMA
worked. It seems that DMA fails for a different reason on each kernel.
On pre1 the error message is
  PCI: Device 00:1f.1 not available because of resource collisions
while on pre2, it's
  PCI: Unable to reserve I/O region #1:8@0 for device 00:1f.1


The complete IDE boot messages:
On pre1:

ide: Assuming 33MHz system bus speed for PIO modes; override with
idebus=xx
ICH4: IDE controller on PCI bus 00 dev f9
PCI: Device 00:1f.1 not available because of resource collisions
ICH4: (ide_setup_pci_device:) Could not enable device.
hda: C/H/S=19161/16/255 from BIOS ignored
hda: MAXTOR 6L040J2, ATA DISK drive
hdb: IC35L060AVVA07-0, ATA DISK drive
hdc: DVDROM 8X, ATAPI CD/DVD-ROM drive
hdd: Memorex CDRW-2216, ATAPI CD/DVD-ROM drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide1 at 0x170-0x177,0x376 on irq 15
hda: 78177792 sectors (40027 MB) w/1819KiB Cache, CHS=77557/16/63
hdb: 120103200 sectors (61493 MB) w/1863KiB Cache, CHS=7476/255/63
hdc: ATAPI DVD-ROM drive, 512kB Cache
Uniform CD-ROM driver Revision: 3.12
hdd: ATAPI 16X CD-ROM CD-R/RW drive, 1024kB Cache
Partition check:
 hda: hda1 hda2 hda3 hda4 < hda5 hda6 >
 hdb: hdb1 hdb2 hdb3 hdb4 
  

On pre2-ac4:
Uniform Multi-Platform E-IDE driver Revision: 6.31
ide: Assuming 33MHz system bus speed for PIO modes; override with
idebus=xx
PCI: Unable to reserve I/O region #1:8@0 for device 00:1f.1
Trying to free nonexistent resource <00000000-00000007>
Trying to free nonexistent resource <00000000-00000003>
Trying to free nonexistent resource <00000000-00000007>
Trying to free nonexistent resource <00000000-00000003>
Trying to free nonexistent resource <0000f000-0000f00f>
Trying to free nonexistent resource <1f800000-1f8003ff>
hda: C/H/S=19161/16/255 from BIOS ignored
hda: MAXTOR 6L040J2, ATA DISK drive
hdb: IC35L060AVVA07-0, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
hdc: DVDROM 8X, ATAPI CD/DVD-ROM drive
hdd: Memorex CDRW-2216, ATAPI CD/DVD-ROM drive
ide1 at 0x170-0x177,0x376 on irq 15
hda: host protected area => 1
hda: 78177792 sectors (40027 MB) w/1819KiB Cache, CHS=77557/16/63
hdb: host protected area => 1
hdb: 120103200 sectors (61493 MB) w/1863KiB Cache, CHS=7476/255/63
hdc: ATAPI DVD-ROM drive, 512kB Cache
Uniform CD-ROM driver Revision: 3.12
hdd: ATAPI 16X CD-ROM CD-R/RW drive, 1024kB Cache
Partition check:
 hda: hda1 hda2 hda3 hda4 < hda5 hda6 >
 hdb: hdb1 hdb2 hdb3 hdb4

Is there any other information I should provide?
-- 
Rob Speer

Re: 2.4.20-pre2-ac4 IDE is slow

[Athanasius]

[-- Attachment #1: Type: text/plain, Size: 1948 bytes --]

On Thu, Aug 22, 2002 at 01:59:45PM -0400, Rob Speer wrote:
> I'm going from 2.4.19 to 2.4.20-pre2-ac4 and the hard drive is noticably
> slower in the new version. (It doesn't use DMA in either version - I
> wish it did in ac4, but that's a separate problem.)
> 
> What I seem to remember from the other message is that there's some
> parameter that can be changed to bring the speed back up. Could someone
> tell me what it is?
> 
> 
> If it helps: output of hdparm /dev/hda
> 
> /dev/hda:
>  multcount    =  0 (off)
>  IO_support   =  0 (default 16-bit)

  This could be part of the problem too?  I don't think I set anything
specific to get to 32bit on my system.  Hmmm, the multcount too?
I'm starting to think, if this is your output from 2.4.20-pre2-ac4 that
you're hitting a bug in the current code which is REALLY not doing the
right thing for your controller.

>  unmaskirq    =  0 (off)
   ^^^^^^^^^^^^^^^^^^^^^^^

  Try hdparm -u1 /dev/hda

>  using_dma    =  0 (off)

  And may as well use -d1 too if it works on the device (yes, I know you
said neither kernel enabled it per default).  Then you're after finding
which DMA mode the IDE bus and drive can do, 'man hdparm' for the
possible values (like -X69 if it's all ATA100 capable for UDMA5).
  With a 60GB Maxtor and an 80pin cable I use:

	/sbin/hdparm -u1 -d1 -X69 /dev/hda

root@emelia:~# hdparm -t -T /dev/hda

/dev/hda:
 Timing buffer-cache reads:   128 MB in  0.52 seconds =246.15 MB/sec
 Timing buffered disk reads:  64 MB in  2.03 seconds = 31.53 MB/sec

Although I'm wondering if with ATA100 and UDMA5 I should be seeing even
better than that.

HTH,

-Ath
-- 
- Athanasius = Athanasius(at)miggy.org.uk / http://www.clan-lovely.org/~athan/
                  Finger athan(at)fysh.org for PGP key
	   "And it's me who is my enemy. Me who beats me up.
Me who makes the monsters. Me who strips my confidence." Paula Cole - ME

[-- Attachment #2: Type: application/pgp-signature, Size: 240 bytes --]