Skip to contentSkip to conversation

DMA blues...System lockup on setting DMA mode using hdparam

DISCUSSION Jasmeet Sidhu 3 participants 3 messages Feb 23, 2001 – Feb 24, 2001

lore
Share & tools
lore
#0 lore raw

Hey guys,

I have five Promise ATA100 controllers configured using kernel version 2.4.2-ac1 (using pdc202xx drivers of course) on ASUS A7V with a AMD Tbird 1GHz processor. Now for the most part this kernel is very stable. I have premium cables connected to the hard drives and all drives in the system are masters as you can probably tell by the drive device letters assigned. The cables are 80pin UDMA (100% Data Integrity). I have not seen any CRC errors, in fact the system has been up overnight and has almost transferred about 105GB of data in various file sizes.

The problem:

when trying to set the DMA mode on the drives, using "hdparm -X69 /dev/hda", it works fine. As a matter of fact, this command succeeds for the following devices:
/dev/hda, /dev/hdc, /dev/hdm, /dev/hdo, /dev/hdq, /dev/hds
However, the system locks up completely when I try the same exact command for *any* of the following devices: /dev/hde, /dev/hdg, /dev/hdi, /dev/hdk.

*NOTE* the Raid5 array /dev/md0 is not running when I am trying to set the DMA modes. The raid is not mounted and is in stopped mode using raidstop /dev/md0.

Also, when I try and use the -k and the -K switches (keep settings after reset), the programs says that it worked. However, after I restart the system, these "flags" are set to 0 again. Is this normal? In other words:
hdparam -k /dev/hda
keepsettings = 0 (off)

# now lets set the -k option (keep settings after refresh).
hdparam -k1 /dev/hda
 setting keep_settings to 1 (on)
 keepsettings =  1 (on)
# noe lets restart the system and query again
hdparam -k /dev/hda
 keepsettings =  0 (off)

Is this normal?

Also another question related to IDE:
Is there anyway we can see how good/bad the system performance is while
the system is working? I am not talking about a benchmarking tool like bonnie that simply tries to figure out how good a system can perform. I would like to see something, maybe in /proc/ide/, that shows me the current throughput of the ide subsystem. For example how many kb of data is going in, how much coming out of each device. Any ideas on how to go about maybe adding this? Where would be an ideal place to add such functionality? In the ide code or maybe in the raid section? Or maybe these two should be kept separate. Any thoughts guys?

Any additional required information can be posted, let me know.

Anybody else out there with a similar situation? Your thoughts on this would be really appreciated.

Here's the setup:

ide0 at 0x3800-0x3807,0x3402 on irq 11 PDC20265
ide1 at 0x3000-0x3007,0x2802 on irq 11
ide2 at 0x5400-0x5407,0x5002 on irq 15 PDC20267
ide3 at 0x4800-0x4807,0x4402 on irq 15
ide4 at 0x7000-0x7007,0x6802 on irq 11 PDC20267
ide5 at 0x6400-0x6407,0x6002 on irq 11
ide6 at 0x8800-0x8807,0x8402 on irq 14 PDC20267
ide7 at 0x8000-0x8007,0x7802 on irq 14
ide8 at 0xa400-0xa407,0xa002 on irq 10 PDC20267
ide9 at 0x9800-0x9807,0x9402 on irq 10

hda: 40188960 sectors (20577 MB) w/1916KiB Cache, CHS=39870/16/63, UDMA(100)
hdc: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hde: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdg: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdi: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdk: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdm: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdo: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdq: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hds: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)

# Raid-5 configuration
#
raiddev /dev/md0
raid-level 5
chunk-size 4
parity-algorithm left-symmetric
persistent-superblock 1
nr-raid-disks 8
nr-spare-disks 1

device          		/dev/hde1
raid-disk       		0
device          		/dev/hdg1
raid-disk       		1
device          		/dev/hdi1
raid-disk       		2
device          		/dev/hdk1
raid-disk       		3
device          		/dev/hdm1
raid-disk       		4
device          		/dev/hdo1
raid-disk       		5
device          		/dev/hdq1
raid-disk      		 6
device          		/dev/hds1
raid-disk       		7
device          		/dev/hdc1
spare-disk      		0

[root@bertha hdparm-3.9]# df -k
Filesystem 1k-blocks Used Available Use% Mounted on
/dev/hda3 19072868 3589612 14514392 20% /
/dev/hda1 198313 11667 176392 7% /boot
/dev/md0 525461076 108657156 416803920 21% /raid
sj-f760-1:/vol/vol03/data01

                     142993408 140675648   2317760  99% /mnt/netapps
#1 lore raw
Quoted 3 lines · Jasmeet Sidhu

Also another question related to IDE:
Is there anyway we can see how good/bad the system performance is while
the system is working?

that information (what the disk subsytem as a whole is doing) is collected in /proc/stat something like xosview can help you visualize it.

Quoted 9 lines

I am not talking about a benchmarking tool like
bonnie that simply tries to figure out how good a system can perform. I would like to see something, maybe in /proc/ide/, that shows me the current throughput of the ide subsystem. For example how many kb of data is going in, how much coming out of each device. Any ideas on how to go about maybe adding this? Where would be an ideal place to add such functionality? In the ide code or maybe in the raid section? Or maybe these two should be kept separate. Any thoughts guys?

Quoted 4 lines

Any additional required information can be posted, let me know.

Anybody else out there with a similar situation? Your thoughts on this would be really appreciated.

I actually have a similar situation although without hanging, with three controllers. the kernels detects the two devices on the third controller as udma rather than udma(100) with 2.4.1.

hda: 30003120 sectors (15362 MB) w/1916KiB Cache, CHS=1867/255/63, UDMA(33)
hdc: 30003120 sectors (15362 MB) w/1916KiB Cache, CHS=29765/16/63, UDMA(33)
first two are on the server-works chipset controller

hde: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdg: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdi: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdk: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdm: 30003120 sectors (15362 MB) w/1916KiB Cache, CHS=29765/16/63, (U)DMA
hdo: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, (U)DMA
these 6 are on three promise contollers... all with 18" 80pin cable

hdd: ATAPI 40X CD-ROM drive, 128kB Cache, UDMA(33)

Quoted 66 lines

Here's the setup:

ide0 at 0x3800-0x3807,0x3402 on irq 11 PDC20265
ide1 at 0x3000-0x3007,0x2802 on irq 11
ide2 at 0x5400-0x5407,0x5002 on irq 15 PDC20267
ide3 at 0x4800-0x4807,0x4402 on irq 15
ide4 at 0x7000-0x7007,0x6802 on irq 11 PDC20267
ide5 at 0x6400-0x6407,0x6002 on irq 11
ide6 at 0x8800-0x8807,0x8402 on irq 14 PDC20267
ide7 at 0x8000-0x8007,0x7802 on irq 14
ide8 at 0xa400-0xa407,0xa002 on irq 10 PDC20267
ide9 at 0x9800-0x9807,0x9402 on irq 10

hda: 40188960 sectors (20577 MB) w/1916KiB Cache, CHS=39870/16/63, UDMA(100)
hdc: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hde: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdg: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdi: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdk: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdm: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdo: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hdq: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)
hds: 150136560 sectors (76870 MB) w/1916KiB Cache, CHS=148945/16/63, UDMA(100)

# Raid-5 configuration
#
raiddev /dev/md0
raid-level 5
chunk-size 4
parity-algorithm left-symmetric
persistent-superblock 1
nr-raid-disks 8
nr-spare-disks 1

device          		/dev/hde1
raid-disk       		0
device          		/dev/hdg1
raid-disk       		1
device          		/dev/hdi1
raid-disk       		2
device          		/dev/hdk1
raid-disk       		3
device          		/dev/hdm1
raid-disk       		4
device          		/dev/hdo1
raid-disk       		5
device          		/dev/hdq1
raid-disk      		 6
device          		/dev/hds1
raid-disk       		7
device          		/dev/hdc1
spare-disk      		0

[root@bertha hdparm-3.9]# df -k
Filesystem 1k-blocks Used Available Use% Mounted on
/dev/hda3 19072868 3589612 14514392 20% /
/dev/hda1 198313 11667 176392 7% /boot
/dev/md0 525461076 108657156 416803920 21% /raid
sj-f760-1:/vol/vol03/data01

                      142993408 140675648   2317760  99% /mnt/netapps

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 8 lines
--------------------------------------------------------------------------
Joel Jaeggli				       [email protected]
Academic User Services			     [email protected]
     PGP Key Fingerprint: 1DE9 8FCA 51FB 4195 B42A 9C32 A30D 121E
--------------------------------------------------------------------------
It is clear that the arm of criticism cannot replace the criticism of
arms.  Karl Marx -- Introduction to the critique of Hegel's Philosophy of
the right, 1843.
#2 lore raw
In reply to Jasmeet Sidhu
Quoted 13 lines · Jasmeet Sidhu

Also, when I try and use the -k and the -K switches (keep settings after reset), the programs says that it worked. However, after I restart the system, these "flags" are set to 0 again. Is this normal? In other words:
hdparam -k /dev/hda
keepsettings = 0 (off)

# now lets set the -k option (keep settings after refresh).
hdparam -k1 /dev/hda
  setting keep_settings to 1 (on)
  keepsettings =  1 (on)
# noe lets restart the system and query again
hdparam -k /dev/hda
  keepsettings =  0 (off)
Quoted 1 line

Is this normal?

This only relates to a ide bus reset in case of a failure, not system reset.

Signature · 2 lines
Vojtech Pavlik
SuSE Labs