DECStation: Support for PMAZ-AA TC SCSI card?

DECStation: Support for PMAZ-AA TC SCSI card?

[Houten K.H.C. van (Karel)]

Hi all,

I'm currently experimenting with software raid support on my decstation,
and it looks fine! But I would love to use more than one SCSI chain
for my raid disks. My DECStation contains a Turbochannel PMAZ-AA
SCSI card, which WAS once supported in the driver, but isn't anymore. :-(

Does anyone knows about patches to get this working again? (Harald, David, 
Florian?)

Regards,
Karel.
-- 
Karel van Houten

----------------------------------------------------------
The box said "Requires Windows 95 or better."
I can't understand why it won't work on my Linux computer. 
----------------------------------------------------------

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Maciej W. Rozycki]

On Mon, 15 Jul 2002, Houten K.H.C. van (Karel) wrote:

> I'm currently experimenting with software raid support on my decstation,
> and it looks fine! But I would love to use more than one SCSI chain
> for my raid disks. My DECStation contains a Turbochannel PMAZ-AA
> SCSI card, which WAS once supported in the driver, but isn't anymore. :-(

 That's basically the same as the /200's onboard SCSI.  If that works, why
wouldn't an additional card (yup, I know the SCSI driver is a mess...)?

 [Looking at the sources...]  The driver seems to have all necessary bits
to support additional HBAs.  What do you mean by "not supported anymore?" 
What does it report for PMAZ-AA cards?

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Karel van Houten]

Hi Maciej,

You wrote:
> 
> On Mon, 15 Jul 2002, Houten K.H.C. van (Karel) wrote:
> 
> > I'm currently experimenting with software raid support on my decstation,
> > and it looks fine! But I would love to use more than one SCSI chain
> > for my raid disks. My DECStation contains a Turbochannel PMAZ-AA
> > SCSI card, which WAS once supported in the driver, but isn't anymore. :-(
> 
>  That's basically the same as the /200's onboard SCSI.  If that works, why
> wouldn't an additional card (yup, I know the SCSI driver is a mess...)?
> 
>  [Looking at the sources...]  The driver seems to have all necessary bits
> to support additional HBAs.  What do you mean by "not supported anymore?" 
> What does it report for PMAZ-AA cards?

Usually I get SCSI bus problems when using the second chain.
Even with devices that don't give any problems when connected to
the on-board bus. Here is my boot log, with the scsi errors.
In this case, I could use the disk on esp1, but I don't know
if I can trust this...

KN05 V2.1k
>>cnfg
 3: KN05     DEC      V2.1k    TCF0  (256 MB)
                                     (enet: 08-00-2b-37-63-76)
                                     (SCSI = 7)
 1: PMAZ-AA  DEC      V5.3d    TCF0  (SCSI = 7)
>>cnfg 1
 1: PMAZ-AA  DEC      V5.3d    TCF0  (SCSI = 7)
    ---------------------------------------------------
    DEV   PID                VID        REV    SCSI DEV
    ===== ================== ========== ====== ========
    rz1   C2490A             COMPAQ     5193   DIR

>>cnfg 3
 3: KN05     DEC      V2.1k    TCF0  (256 MB)
                                     (enet: 08-00-2b-37-63-76)
                                     (SCSI = 7)
            ---------------------------------------------------
            DEV   PID                VID        REV    SCSI DEV
            ===== ================== ========== ====== ========
            rz0   DCHS04U            IBM        6464   DIR
            rz1   C2490A             COMPAQ     5193   DIR
            rz2   C2490A             COMPAQ     5193   DIR
            rz3   C2490A             COMPAQ     5193   DIR
            rz4   C2490A             COMPAQ     5193   DIR
            rz5   C2490A             COMPAQ     5193   DIR
            rz6   CD-ROM CDU-55S     SONY       1.0t   CD-ROM

        cache: I(16 KB), D(16 KB), S(1024 KB);  Scache line (32 bytes)
        processor revision (4.0)
        mem( 0):  a0000000:a1ffffff  ( 32 MB)
        mem( 1):  a2000000:a3ffffff  ( 32 MB)
        mem( 2):  a4000000:a5ffffff  ( 32 MB)
        mem( 3):  a6000000:a7ffffff  ( 32 MB)
        mem( 4):  a8000000:a9ffffff  ( 32 MB)
        mem( 5):  aa000000:abffffff  ( 32 MB)
        mem( 6):  ac000000:adffffff  ( 32 MB)
        mem( 7):  ae000000:afffffff  ( 32 MB)

>>boot
delo V0.7 Copyright 2000 Florian Lohoff <flo@rfc822.org>
Loading /etc/delo.conf .. ok
Loading /boot/vmlinux-2.4.18 ....... ok
This DECstation is a DS5000/2x0
CPU revision is: 00000440
FPU revision is: 00000500
Primary instruction cache 16kb, linesize 16 bytes.
Primary data cache 16kb, linesize 16 bytes.
Secondary cache sized at 1024K linesize 32 bytes.
Linux version 2.4.18 (root@elrond) (gcc version 2.96 20000731 (Red Hat Linux 7.3 2.96-110.1)) #12 Mon Jul 1 18:07:40 MEST 2002
Determined physical RAM map:
 memory: 10000000 @ 00000000 (usable)
On node 0 totalpages: 65536
zone(0): 65536 pages.
zone(1): 0 pages.
zone(2): 0 pages.
Kernel command line: root=/dev/sda1 console=ttyS2 ro
Calibrating delay loop... 59.86 BogoMIPS
Memory: 255572k/262144k available (1804k kernel code, 6572k reserved, 108k data, 76k init, 0k highmem)
Dentry-cache hash table entries: 32768 (order: 6, 262144 bytes)
Inode-cache hash table entries: 16384 (order: 5, 131072 bytes)
Mount-cache hash table entries: 4096 (order: 3, 32768 bytes)
Buffer-cache hash table entries: 16384 (order: 4, 65536 bytes)
Page-cache hash table entries: 65536 (order: 6, 262144 bytes)
Checking for 'wait' instruction...  unavailable.
POSIX conformance testing by UNIFIX
TURBOchannel rev. 1 at 25.0 MHz (without parity)
    slot 1: DEC      PMAZ-AA  V5.3d
Linux NET4.0 for Linux 2.4
Based upon Swansea University Computer Society NET3.039
Initializing RT netlink socket
Starting kswapd
Journalled Block Device driver loaded
devfs: v1.10 (20020120) Richard Gooch (rgooch@atnf.csiro.au)
devfs: boot_options: 0x0
lk201: DECstation LK keyboard driver v0.05.
pty: 256 Unix98 ptys configured
DECstation Z8530 serial driver version 0.07
ttyS00 at 0xbf900001 (irq = 14) is a Z85C30 SCC
ttyS01 at 0xbf900009 (irq = 14) is a Z85C30 SCC
ttyS02 at 0xbf980001 (irq = 15) is a Z85C30 SCC
rtc: Digital DECstation epoch (2000) detected
Real Time Clock Driver v1.10e
block: 128 slots per queue, batch=32
declance.c: v0.009 by Linux MIPS DECstation task force
eth0: IOASIC onboard LANCE, addr = 08:00:2b:37:63:76, irq = 16
SCSI subsystem driver Revision: 1.00
SCSI ID 7 Clk 25MHz CCF=5 TOut 167 NCR53C9x(esp236)
SCSI ID 7 Clk 25MHz CCF=5 TOut 167 NCR53C9x(esp236)
ESP: Total of 2 ESP hosts found, 2 actually in use.
scsi0 : ESP236 (NCR53C9x)
scsi1 : ESP236 (NCR53C9x)
esp0: AIEEE wide msg received
esp0: hoping for msgout
  Vendor: IBM       Model: DCHS04U           Rev: 6464
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: SONY      Model: CD-ROM CDU-55S    Rev: 1.0t
  Type:   CD-ROM                             ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
Attached scsi disk sdb at scsi0, channel 0, id 1, lun 0
Attached scsi disk sdc at scsi0, channel 0, id 2, lun 0
Attached scsi disk sdd at scsi0, channel 0, id 3, lun 0
Attached scsi disk sde at scsi0, channel 0, id 4, lun 0
Attached scsi disk sdf at scsi0, channel 0, id 5, lun 0
Attached scsi disk sdg at scsi1, channel 0, id 1, lun 0
SCSI device sda: 8813870 512-byte hdwr sectors (4513 MB)
Partition check:
 /dev/scsi/host0/bus0/target0/lun0: p1 p2 p3
esp0: target 1 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdb: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target1/lun0: p1
esp0: target 2 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdc: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target2/lun0: p1
esp0: target 3 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdd: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target3/lun0: p1
esp0: target 4 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sde: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target4/lun0: p1
esp0: target 5 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdf: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target5/lun0: p1
scsi : aborting command due to timeout : pid 52, scsi1, channel 0, id 1, lun 0 Request Sense 00 00 00 40 00
esp1: Aborting command
esp1: dumping state
esp1: SW [sreg<00> sstep<04> ireg<20>]
esp1: HW reread [sreg<06> sstep<c1> ireg<00>]
esp1: current command [tgt<01> lun<00> pphase<UNISSUED> cphase<SLCTMSG>]
esp1: disconnected
SCSI host 1 abort (pid 52) timed out - resetting
SCSI bus is being reset for host 1 channel 0.
esp1: Resetting scsi bus
esp1: Gross error sreg=40
esp1: SCSI bus reset interrupt
esp1: Warning, live target 1 not responding to selection.
esp1: Warning, live target 1 not responding to selection.
SCSI device sdg: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host1/bus0/target1/lun0: p1 p2
Attached scsi CD-ROM sr0 at scsi0, channel 0, id 6, lun 0
esp0: target 6 [period 248ns offset 15 4.03MHz synchronous SCSI]
sr0: scsi-1 drive
Uniform CD-ROM driver Revision: 3.12
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP, IGMP
IP: routing cache hash table of 2048 buckets, 16Kbytes
TCP: Hash tables configured (established 16384 bind 16384)
NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
kjournald starting.  Commit interval 5 seconds
EXT3-fs: mounted filesystem with ordered data mode.
VFS: Mounted root (ext3 filesystem) readonly.
Freeing unused PROM memory: 124k freed
Freeing unused kernel memory: 76k freed
INIT: version 2.84 booting


Thanks for your time...
Regards,

-- 
Karel van Houten

----------------------------------------------------------
The box said "Requires Windows 95 or better."
I can't understand why it won't work on my Linux computer. 
----------------------------------------------------------

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Karel van Houten]

Hi Maciej,

You wrote:
> 
> On Mon, 15 Jul 2002, Houten K.H.C. van (Karel) wrote:
> 
> > I'm currently experimenting with software raid support on my decstation,
> > and it looks fine! But I would love to use more than one SCSI chain
> > for my raid disks. My DECStation contains a Turbochannel PMAZ-AA
> > SCSI card, which WAS once supported in the driver, but isn't anymore. :-(
> 
>  That's basically the same as the /200's onboard SCSI.  If that works, why
> wouldn't an additional card (yup, I know the SCSI driver is a mess...)?
> 
>  [Looking at the sources...]  The driver seems to have all necessary bits
> to support additional HBAs.  What do you mean by "not supported anymore?" 
> What does it report for PMAZ-AA cards?

Usually I get SCSI bus problems when using the second chain.
Even with devices that don't give any problems when connected to
the on-board bus. Here is my boot log, with the scsi errors.
In this case, I could use the disk on esp1, but I don't know
if I can trust this...

KN05 V2.1k
>>cnfg
 3: KN05     DEC      V2.1k    TCF0  (256 MB)
                                     (enet: 08-00-2b-37-63-76)
                                     (SCSI = 7)
 1: PMAZ-AA  DEC      V5.3d    TCF0  (SCSI = 7)
>>cnfg 1
 1: PMAZ-AA  DEC      V5.3d    TCF0  (SCSI = 7)
    ---------------------------------------------------
    DEV   PID                VID        REV    SCSI DEV
    ===== ================== ========== ====== ========
    rz1   C2490A             COMPAQ     5193   DIR

>>cnfg 3
 3: KN05     DEC      V2.1k    TCF0  (256 MB)
                                     (enet: 08-00-2b-37-63-76)
                                     (SCSI = 7)
            ---------------------------------------------------
            DEV   PID                VID        REV    SCSI DEV
            ===== ================== ========== ====== ========
            rz0   DCHS04U            IBM        6464   DIR
            rz1   C2490A             COMPAQ     5193   DIR
            rz2   C2490A             COMPAQ     5193   DIR
            rz3   C2490A             COMPAQ     5193   DIR
            rz4   C2490A             COMPAQ     5193   DIR
            rz5   C2490A             COMPAQ     5193   DIR
            rz6   CD-ROM CDU-55S     SONY       1.0t   CD-ROM

        cache: I(16 KB), D(16 KB), S(1024 KB);  Scache line (32 bytes)
        processor revision (4.0)
        mem( 0):  a0000000:a1ffffff  ( 32 MB)
        mem( 1):  a2000000:a3ffffff  ( 32 MB)
        mem( 2):  a4000000:a5ffffff  ( 32 MB)
        mem( 3):  a6000000:a7ffffff  ( 32 MB)
        mem( 4):  a8000000:a9ffffff  ( 32 MB)
        mem( 5):  aa000000:abffffff  ( 32 MB)
        mem( 6):  ac000000:adffffff  ( 32 MB)
        mem( 7):  ae000000:afffffff  ( 32 MB)

>>boot
delo V0.7 Copyright 2000 Florian Lohoff <flo@rfc822.org>
Loading /etc/delo.conf .. ok
Loading /boot/vmlinux-2.4.18 ....... ok
This DECstation is a DS5000/2x0
CPU revision is: 00000440
FPU revision is: 00000500
Primary instruction cache 16kb, linesize 16 bytes.
Primary data cache 16kb, linesize 16 bytes.
Secondary cache sized at 1024K linesize 32 bytes.
Linux version 2.4.18 (root@elrond) (gcc version 2.96 20000731 (Red Hat Linux 7.3 2.96-110.1)) #12 Mon Jul 1 18:07:40 MEST 2002
Determined physical RAM map:
 memory: 10000000 @ 00000000 (usable)
On node 0 totalpages: 65536
zone(0): 65536 pages.
zone(1): 0 pages.
zone(2): 0 pages.
Kernel command line: root=/dev/sda1 console=ttyS2 ro
Calibrating delay loop... 59.86 BogoMIPS
Memory: 255572k/262144k available (1804k kernel code, 6572k reserved, 108k data, 76k init, 0k highmem)
Dentry-cache hash table entries: 32768 (order: 6, 262144 bytes)
Inode-cache hash table entries: 16384 (order: 5, 131072 bytes)
Mount-cache hash table entries: 4096 (order: 3, 32768 bytes)
Buffer-cache hash table entries: 16384 (order: 4, 65536 bytes)
Page-cache hash table entries: 65536 (order: 6, 262144 bytes)
Checking for 'wait' instruction...  unavailable.
POSIX conformance testing by UNIFIX
TURBOchannel rev. 1 at 25.0 MHz (without parity)
    slot 1: DEC      PMAZ-AA  V5.3d
Linux NET4.0 for Linux 2.4
Based upon Swansea University Computer Society NET3.039
Initializing RT netlink socket
Starting kswapd
Journalled Block Device driver loaded
devfs: v1.10 (20020120) Richard Gooch (rgooch@atnf.csiro.au)
devfs: boot_options: 0x0
lk201: DECstation LK keyboard driver v0.05.
pty: 256 Unix98 ptys configured
DECstation Z8530 serial driver version 0.07
ttyS00 at 0xbf900001 (irq = 14) is a Z85C30 SCC
ttyS01 at 0xbf900009 (irq = 14) is a Z85C30 SCC
ttyS02 at 0xbf980001 (irq = 15) is a Z85C30 SCC
rtc: Digital DECstation epoch (2000) detected
Real Time Clock Driver v1.10e
block: 128 slots per queue, batch=32
declance.c: v0.009 by Linux MIPS DECstation task force
eth0: IOASIC onboard LANCE, addr = 08:00:2b:37:63:76, irq = 16
SCSI subsystem driver Revision: 1.00
SCSI ID 7 Clk 25MHz CCF=5 TOut 167 NCR53C9x(esp236)
SCSI ID 7 Clk 25MHz CCF=5 TOut 167 NCR53C9x(esp236)
ESP: Total of 2 ESP hosts found, 2 actually in use.
scsi0 : ESP236 (NCR53C9x)
scsi1 : ESP236 (NCR53C9x)
esp0: AIEEE wide msg received
esp0: hoping for msgout
  Vendor: IBM       Model: DCHS04U           Rev: 6464
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: SONY      Model: CD-ROM CDU-55S    Rev: 1.0t
  Type:   CD-ROM                             ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
Attached scsi disk sdb at scsi0, channel 0, id 1, lun 0
Attached scsi disk sdc at scsi0, channel 0, id 2, lun 0
Attached scsi disk sdd at scsi0, channel 0, id 3, lun 0
Attached scsi disk sde at scsi0, channel 0, id 4, lun 0
Attached scsi disk sdf at scsi0, channel 0, id 5, lun 0
Attached scsi disk sdg at scsi1, channel 0, id 1, lun 0
SCSI device sda: 8813870 512-byte hdwr sectors (4513 MB)
Partition check:
 /dev/scsi/host0/bus0/target0/lun0: p1 p2 p3
esp0: target 1 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdb: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target1/lun0: p1
esp0: target 2 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdc: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target2/lun0: p1
esp0: target 3 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdd: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target3/lun0: p1
esp0: target 4 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sde: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target4/lun0: p1
esp0: target 5 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdf: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target5/lun0: p1
scsi : aborting command due to timeout : pid 52, scsi1, channel 0, id 1, lun 0 Request Sense 00 00 00 40 00
esp1: Aborting command
esp1: dumping state
esp1: SW [sreg<00> sstep<04> ireg<20>]
esp1: HW reread [sreg<06> sstep<c1> ireg<00>]
esp1: current command [tgt<01> lun<00> pphase<UNISSUED> cphase<SLCTMSG>]
esp1: disconnected
SCSI host 1 abort (pid 52) timed out - resetting
SCSI bus is being reset for host 1 channel 0.
esp1: Resetting scsi bus
esp1: Gross error sreg=40
esp1: SCSI bus reset interrupt
esp1: Warning, live target 1 not responding to selection.
esp1: Warning, live target 1 not responding to selection.
SCSI device sdg: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host1/bus0/target1/lun0: p1 p2
Attached scsi CD-ROM sr0 at scsi0, channel 0, id 6, lun 0
esp0: target 6 [period 248ns offset 15 4.03MHz synchronous SCSI]
sr0: scsi-1 drive
Uniform CD-ROM driver Revision: 3.12
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP, IGMP
IP: routing cache hash table of 2048 buckets, 16Kbytes
TCP: Hash tables configured (established 16384 bind 16384)
NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
kjournald starting.  Commit interval 5 seconds
EXT3-fs: mounted filesystem with ordered data mode.
VFS: Mounted root (ext3 filesystem) readonly.
Freeing unused PROM memory: 124k freed
Freeing unused kernel memory: 76k freed
INIT: version 2.84 booting


Thanks for your time...
Regards,

-- 
Karel van Houten

----------------------------------------------------------
The box said "Requires Windows 95 or better."
I can't understand why it won't work on my Linux computer. 
----------------------------------------------------------

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Maciej W. Rozycki]

Hi Karel,

> Usually I get SCSI bus problems when using the second chain.
> Even with devices that don't give any problems when connected to
> the on-board bus. Here is my boot log, with the scsi errors.
> In this case, I could use the disk on esp1, but I don't know
> if I can trust this...
> 
> KN05 V2.1k
> >>cnfg
>  3: KN05     DEC      V2.1k    TCF0  (256 MB)
>                                      (enet: 08-00-2b-37-63-76)
>                                      (SCSI = 7)
>  1: PMAZ-AA  DEC      V5.3d    TCF0  (SCSI = 7)

 Well, you have a mixed system, so it's quite possible PMAZ-A support does
not work reliably anywhere.  I don't have such a card, but specs are
available and the support code is about one screen long.  So it should be
fairly trivial to verify -- I'll look at it. 

 Also you have a KN05 system, which doesn't help, unfortunately.  The KN05
module implements aggressive posting of uncached (read: iomem) writes (see
also /proc/interrupts on your system) and synchronization primitives are
non-existent.  Since for half a year there is no agreement on how generic
synchronization should look like for MIPS, I'm more and more tempted to
add a local hack which at least will let DECstations to perform reliably.
It's quite possible the lack of synchronization is the lone reason of your
problems. 

  Maciej

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Houten K.H.C. van (Karel)]

Hi Maciej,


> > KN05 V2.1k
> > >>cnfg
> >  3: KN05     DEC      V2.1k    TCF0  (256 MB)
> >                                      (enet: 08-00-2b-37-63-76)
> >                                      (SCSI = 7)
> >  1: PMAZ-AA  DEC      V5.3d    TCF0  (SCSI = 7)
> 
>  Well, you have a mixed system, so it's quite possible PMAZ-A support does
> not work reliably anywhere.  I don't have such a card, but specs are
> available and the support code is about one screen long.  So it should be
> fairly trivial to verify -- I'll look at it. 
What do you mean by a mixed system?
 
>  Also you have a KN05 system, which doesn't help, unfortunately.  The KN05
> module implements aggressive posting of uncached (read: iomem) writes (see
> also /proc/interrupts on your system) and synchronization primitives are
> non-existent.  Since for half a year there is no agreement on how generic
> synchronization should look like for MIPS, I'm more and more tempted to
> add a local hack which at least will let DECstations to perform reliably.
> It's quite possible the lack of synchronization is the lone reason of your
> problems. 

Well, if there is anything I could do to help you by testing things,
I'm eager to do so. 

I also have a 5000/200 system. Would it be interesting to put the
PMAZ-AA into that system, and see how it behaves?

Regards,
Karel.


-- 
Karel van Houten

----------------------------------------------------------
The box said "Requires Windows 95 or better."
I can't understand why it won't work on my Linux computer. 
----------------------------------------------------------

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Maciej W. Rozycki]

Hi Karel,

> What do you mean by a mixed system?

 Both a PMAZ-A and an I/O ASIC-based controller.  They differ.

> I also have a 5000/200 system. Would it be interesting to put the
> PMAZ-AA into that system, and see how it behaves?

 Well, the /200's onboard HBA is identical to a PMAZ-A.  You don't need to
rearrange hardware, although you may, just to be sure -- if the /200
works, then the problem is almost surely related to write posting
implemented on the KN05 module which we currently don't handle (it's
worked around in the driver in the I/O ASIC-related functions via an ugly
hack). 

 Just in case I am right, please check if the following hack helps with
your PMAZ-A in your /260. 

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

patch-mips-2.4.19-rc1-20020717-dec_esp-test-0
diff -up --recursive --new-file linux-mips-2.4.19-rc1-20020717.macro/drivers/scsi/dec_esp.c linux-mips-2.4.19-rc1-20020717/drivers/scsi/dec_esp.c
--- linux-mips-2.4.19-rc1-20020717.macro/drivers/scsi/dec_esp.c	2002-04-10 02:58:49.000000000 +0000
+++ linux-mips-2.4.19-rc1-20020717/drivers/scsi/dec_esp.c	2002-07-17 13:24:59.000000000 +0000
@@ -486,12 +486,15 @@ static void pmaz_dma_drain(struct NCR_ES
 static void pmaz_dma_init_read(struct NCR_ESP *esp, __u32 vaddress, int length)
 {
 	volatile int *dmareg = (volatile int *) (esp->slot + DEC_SCSI_DMAREG);
+	volatile unsigned int *dummy = (volatile unsigned int *)KSEG1;
 
 	if (length > ESP_TGT_DMA_SIZE)
 		length = ESP_TGT_DMA_SIZE;
 
 	*dmareg = TC_ESP_DMA_ADDR(esp->slot + DEC_SCSI_SRAM + ESP_TGT_DMA_SIZE);
 
+	*dummy;
+
 	esp_virt_buffer = vaddress;
 	scsi_current_length = length;
 }
@@ -499,6 +502,7 @@ static void pmaz_dma_init_read(struct NC
 static void pmaz_dma_init_write(struct NCR_ESP *esp, __u32 vaddress, int length)
 {
 	volatile int *dmareg = (volatile int *) ( esp->slot + DEC_SCSI_DMAREG );
+	volatile unsigned int *dummy = (volatile unsigned int *)KSEG1;
 
 	memcpy((void *) (esp->slot + DEC_SCSI_SRAM + ESP_TGT_DMA_SIZE),
 			KSEG0ADDR((void *) vaddress), length);
@@ -506,6 +510,7 @@ static void pmaz_dma_init_write(struct N
 	*dmareg = TC_ESP_DMAR_WRITE | 
 		TC_ESP_DMA_ADDR(esp->slot + DEC_SCSI_SRAM + ESP_TGT_DMA_SIZE);
 
+	*dummy;
 }
 
 static void pmaz_dma_ints_off(struct NCR_ESP *esp)

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Houten K.H.C. van (Karel)]

Hi Maciej,

"Maciej W. Rozycki" writes:
> ...
> Just in case I am right, please check if the following hack helps with
>your PMAZ-A in your /260. 

Sorry, same result. See attached log.

Regards,
Karel.

>>boot 3/rz0 1/new
delo V0.7 Copyright 2000 Florian Lohoff <flo@rfc822.org>
Loading /etc/delo.conf .. ok
Loading /boot/vmlinux-2.4.18-test ....... ok
This DECstation is a DS5000/2x0
CPU revision is: 00000440
FPU revision is: 00000500
Primary instruction cache 16kb, linesize 16 bytes.
Primary data cache 16kb, linesize 16 bytes.
Secondary cache sized at 1024K linesize 32 bytes.
Linux version 2.4.18 (root@elrond) (gcc version 2.96 20000731 (Red Hat Linux 7.3 2.96-110.1)) #13 Wed Jul 17 16:58:38 MEST 2002
Determined physical RAM map:
 memory: 10000000 @ 00000000 (usable)
On node 0 totalpages: 65536
zone(0): 65536 pages.
zone(1): 0 pages.
zone(2): 0 pages.
Kernel command line: root=/dev/sda1 console=ttyS2 ro
Calibrating delay loop... 59.86 BogoMIPS
Memory: 255564k/262144k available (1813k kernel code, 6580k reserved, 108k data, 76k init, 0k highmem)
Dentry-cache hash table entries: 32768 (order: 6, 262144 bytes)
Inode-cache hash table entries: 16384 (order: 5, 131072 bytes)
Mount-cache hash table entries: 4096 (order: 3, 32768 bytes)
Buffer-cache hash table entries: 16384 (order: 4, 65536 bytes)
Page-cache hash table entries: 65536 (order: 6, 262144 bytes)
Checking for 'wait' instruction...  unavailable.
POSIX conformance testing by UNIFIX
TURBOchannel rev. 1 at 25.0 MHz (without parity)
    slot 1: DEC      PMAZ-AA  V5.3d
Linux NET4.0 for Linux 2.4
Based upon Swansea University Computer Society NET3.039
Initializing RT netlink socket
Starting kswapd
Journalled Block Device driver loaded
devfs: v1.10 (20020120) Richard Gooch (rgooch@atnf.csiro.au)
devfs: boot_options: 0x0
lk201: DECstation LK keyboard driver v0.05.
pty: 256 Unix98 ptys configured
DECstation Z8530 serial driver version 0.07
ttyS00 at 0xbf900001 (irq = 14) is a Z85C30 SCC
ttyS01 at 0xbf900009 (irq = 14) is a Z85C30 SCC
ttyS02 at 0xbf980001 (irq = 15) is a Z85C30 SCC
rtc: Digital DECstation epoch (2000) detected
Real Time Clock Driver v1.10e
block: 128 slots per queue, batch=32
declance.c: v0.009 by Linux MIPS DECstation task force
eth0: IOASIC onboard LANCE, addr = 08:00:2b:37:63:76, irq = 16
SCSI subsystem driver Revision: 1.00
SCSI ID 7 Clk 25MHz CCF=5 TOut 167 NCR53C9x(esp236)
SCSI ID 7 Clk 25MHz CCF=5 TOut 167 NCR53C9x(esp236)
ESP: Total of 2 ESP hosts found, 2 actually in use.
scsi0 : ESP236 (NCR53C9x)
scsi1 : ESP236 (NCR53C9x)
esp0: AIEEE wide msg received
esp0: hoping for msgout
  Vendor: IBM       Model: DCHS04U           Rev: 6464
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
  Vendor: SONY      Model: CD-ROM CDU-55S    Rev: 1.0t
  Type:   CD-ROM                             ANSI SCSI revision: 02
  Vendor: COMPAQ    Model: C2490A            Rev: 5193
  Type:   Direct-Access                      ANSI SCSI revision: 02
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
Attached scsi disk sdb at scsi0, channel 0, id 1, lun 0
Attached scsi disk sdc at scsi0, channel 0, id 2, lun 0
Attached scsi disk sdd at scsi0, channel 0, id 3, lun 0
Attached scsi disk sde at scsi0, channel 0, id 4, lun 0
Attached scsi disk sdf at scsi0, channel 0, id 5, lun 0
Attached scsi disk sdg at scsi1, channel 0, id 1, lun 0
SCSI device sda: 8813870 512-byte hdwr sectors (4513 MB)
Partition check:
 /dev/scsi/host0/bus0/target0/lun0: p1 p2 p3
esp0: target 1 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdb: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target1/lun0: p1
esp0: target 2 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdc: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target2/lun0: p1
esp0: target 3 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdd: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target3/lun0: p1
esp0: target 4 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sde: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target4/lun0: p1
esp0: target 5 [period 200ns offset 8 5.00MHz synchronous SCSI]
SCSI device sdf: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host0/bus0/target5/lun0: p1
scsi : aborting command due to timeout : pid 52, scsi1, channel 0, id 1, lun 0 Request Sense 00 00 00 40 00
esp1: Aborting command
esp1: dumping state
esp1: SW [sreg<00> sstep<04> ireg<20>]
esp1: HW reread [sreg<06> sstep<c1> ireg<00>]
esp1: current command [tgt<01> lun<00> pphase<UNISSUED> cphase<SLCTMSG>]
esp1: disconnected
SCSI host 1 abort (pid 52) timed out - resetting
SCSI bus is being reset for host 1 channel 0.
esp1: Resetting scsi bus
esp1: Gross error sreg=40
esp1: SCSI bus reset interrupt
esp1: Warning, live target 1 not responding to selection.
esp1: Warning, live target 1 not responding to selection.
SCSI device sdg: 4110000 512-byte hdwr sectors (2104 MB)
 /dev/scsi/host1/bus0/target1/lun0: p1 p2
Attached scsi CD-ROM sr0 at scsi0, channel 0, id 6, lun 0
esp0: target 6 [period 248ns offset 15 4.03MHz synchronous SCSI]
sr0: scsi-1 drive

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Maciej W. Rozycki]

Hi Karel,

> Sorry, same result. See attached log.

 Thanks for the report.  Apart from missing WB flushing, there is nothing
obviously broken in the PMAZ-A code -- I'll look at the problem more
deeply later.  The driver seems to work to some extent as it was able to
retrieve inquiry data, so it's not broken in principle.

 It would be great if you could check if the driver works for the /200's
onboard PMAZ-A.  If it worked there, I'd suspect a bug in the NCR53C8x
support core.  But please don't put a second PMAZ-A into your /200 -- for
an unclear reason the driver only supports a single I/O ASIC-based HBA and
a single additional PMAZ-A board.  All PMAZ-A boards share operational
variables with one another, so using more than a single one leads to data
corruption.

  Maciej

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Karel van Houten]

Hi Maciej,

> 
>  It would be great if you could check if the driver works for the /200's
> onboard PMAZ-A.  If it worked there, I'd suspect a bug in the NCR53C8x
> support core.  But please don't put a second PMAZ-A into your /200 -- for
> an unclear reason the driver only supports a single I/O ASIC-based HBA and
> a single additional PMAZ-A board.  All PMAZ-A boards share operational
> variables with one another, so using more than a single one leads to data
> corruption.

I have a 5000/200 running fine with the same kernel (the one without
your patch). Or did you mean WITH your patch? The only problem
is that delo can't handle the different prom in the /200, so that
system has to boot over the network, but can use the local disks just fine.

Regards,
-- 
Karel van Houten

----------------------------------------------------------
The box said "Requires Windows 95 or better."
I can't understand why it won't work on my Linux computer. 
----------------------------------------------------------

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Karel van Houten]

Hi Maciej,

> 
>  It would be great if you could check if the driver works for the /200's
> onboard PMAZ-A.  If it worked there, I'd suspect a bug in the NCR53C8x
> support core.  But please don't put a second PMAZ-A into your /200 -- for
> an unclear reason the driver only supports a single I/O ASIC-based HBA and
> a single additional PMAZ-A board.  All PMAZ-A boards share operational
> variables with one another, so using more than a single one leads to data
> corruption.

I have a 5000/200 running fine with the same kernel (the one without
your patch). Or did you mean WITH your patch? The only problem
is that delo can't handle the different prom in the /200, so that
system has to boot over the network, but can use the local disks just fine.

Regards,
-- 
Karel van Houten

----------------------------------------------------------
The box said "Requires Windows 95 or better."
I can't understand why it won't work on my Linux computer. 
----------------------------------------------------------

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Maciej W. Rozycki]

Hi Karel,

> I have a 5000/200 running fine with the same kernel (the one without
> your patch). Or did you mean WITH your patch? The only problem

 If it works without the patch, it will also do with it.

> is that delo can't handle the different prom in the /200, so that
> system has to boot over the network, but can use the local disks just fine.

 OK, more writeback fixes.  Please get the following patches: 

- patch-mips-2.4.18-20020530-mb-wb-8.gz,

- patch-mips-2.4.18-20020625-wbflush-7.gz

from 'ftp://ftp.ds2.pg.gda.pl/pub/macro/linux/' and replace the hack I
sent you yesterday with the following real fix.  After applying the three
patches you need to rebuild the kernel from scratch, i.e. do `make
oldconfig dep clean boot modules' as the two above patches modify the
kernel's configuration.

 Please report if this works or not. 

  Maciej

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

patch-mips-2.4.19-rc1-20020717-dec_esp-test-1
diff -up --recursive --new-file linux-mips-2.4.19-rc1-20020717.macro/drivers/scsi/dec_esp.c linux-mips-2.4.19-rc1-20020717/drivers/scsi/dec_esp.c
--- linux-mips-2.4.19-rc1-20020717.macro/drivers/scsi/dec_esp.c	2002-04-10 02:58:49.000000000 +0000
+++ linux-mips-2.4.19-rc1-20020717/drivers/scsi/dec_esp.c	2002-07-18 16:33:22.000000000 +0000
@@ -492,6 +492,8 @@ static void pmaz_dma_init_read(struct NC
 
 	*dmareg = TC_ESP_DMA_ADDR(esp->slot + DEC_SCSI_SRAM + ESP_TGT_DMA_SIZE);
 
+	iob();
+
 	esp_virt_buffer = vaddress;
 	scsi_current_length = length;
 }
@@ -506,6 +508,7 @@ static void pmaz_dma_init_write(struct N
 	*dmareg = TC_ESP_DMAR_WRITE | 
 		TC_ESP_DMA_ADDR(esp->slot + DEC_SCSI_SRAM + ESP_TGT_DMA_SIZE);
 
+	iob();
 }
 
 static void pmaz_dma_ints_off(struct NCR_ESP *esp)

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Dave Airlie]

Hi,

about 30 mins coding should be able to hack around the two cards in one
system issue :-) (Flo did some work already).

only esp_virt_buffer and  scsi_current_length globals are used for the
PMAZ-A as far as I know, and only in the read path between
pmaz_dma_init_read and pmaz_dma_drain, also maybe the pmaz_cmd_buffer when
I look at it.

if there is nowhere in the esp to place them perhaps a priv void * needs
to be added to the NCR core code and used to store this stuff, I meant to
do it at the time, but twas 2 years ago and at the moment I'm nearly as
far away from my DecStation as physically possible and not getting any
closer for the forseeable :-)

My reason of course was I didn't really know much about TC and that such a
card existed orignally, and the original code only handled one IO-ASIC
(which I think is okay)...

Dave. in Laos.

> Hi Karel,
>
>> Sorry, same result. See attached log.
>
>  Thanks for the report.  Apart from missing WB flushing, there is
> nothing
> obviously broken in the PMAZ-A code -- I'll look at the problem more
> deeply later.  The driver seems to work to some extent as it was able to
> retrieve inquiry data, so it's not broken in principle.
>
>  It would be great if you could check if the driver works for the /200's
> onboard PMAZ-A.  If it worked there, I'd suspect a bug in the NCR53C8x
> support core.  But please don't put a second PMAZ-A into your /200 --
> for an unclear reason the driver only supports a single I/O ASIC-based
> HBA and a single additional PMAZ-A board.  All PMAZ-A boards share
> operational variables with one another, so using more than a single one
> leads to data corruption.
>
>   Maciej
>
> --
> +  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
> +--------------------------------------------------------------+
> +        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Dave Airlie]

Hi,

about 30 mins coding should be able to hack around the two cards in one
system issue :-) (Flo did some work already).

only esp_virt_buffer and  scsi_current_length globals are used for the
PMAZ-A as far as I know, and only in the read path between
pmaz_dma_init_read and pmaz_dma_drain, also maybe the pmaz_cmd_buffer when
I look at it.

if there is nowhere in the esp to place them perhaps a priv void * needs
to be added to the NCR core code and used to store this stuff, I meant to
do it at the time, but twas 2 years ago and at the moment I'm nearly as
far away from my DecStation as physically possible and not getting any
closer for the forseeable :-)

My reason of course was I didn't really know much about TC and that such a
card existed orignally, and the original code only handled one IO-ASIC
(which I think is okay)...

Dave. in Laos.

> Hi Karel,
>
>> Sorry, same result. See attached log.
>
>  Thanks for the report.  Apart from missing WB flushing, there is
> nothing
> obviously broken in the PMAZ-A code -- I'll look at the problem more
> deeply later.  The driver seems to work to some extent as it was able to
> retrieve inquiry data, so it's not broken in principle.
>
>  It would be great if you could check if the driver works for the /200's
> onboard PMAZ-A.  If it worked there, I'd suspect a bug in the NCR53C8x
> support core.  But please don't put a second PMAZ-A into your /200 --
> for an unclear reason the driver only supports a single I/O ASIC-based
> HBA and a single additional PMAZ-A board.  All PMAZ-A boards share
> operational variables with one another, so using more than a single one
> leads to data corruption.
>
>   Maciej
>
> --
> +  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
> +--------------------------------------------------------------+
> +        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Maciej W. Rozycki]

Dave,

> about 30 mins coding should be able to hack around the two cards in one
> system issue :-) (Flo did some work already).

 Of course, if you have them, then don't hesitate to do that.  I'm going
to clean the driver up soon, but it's not a priority for me.

> only esp_virt_buffer and  scsi_current_length globals are used for the
> PMAZ-A as far as I know, and only in the read path between
> pmaz_dma_init_read and pmaz_dma_drain, also maybe the pmaz_cmd_buffer when
> I look at it.

 Exactly.

> if there is nowhere in the esp to place them perhaps a priv void * needs
> to be added to the NCR core code and used to store this stuff, I meant to
> do it at the time, but twas 2 years ago and at the moment I'm nearly as
> far away from my DecStation as physically possible and not getting any
> closer for the forseeable :-)

 No need to, "struct NCR_ESP" provides esp_id which may be used to index
private data.

> My reason of course was I didn't really know much about TC and that such a
> card existed orignally, and the original code only handled one IO-ASIC
> (which I think is okay)...

 Well, existing hardware permits up to six TC cards to be put into a
single system.  That's the limit for Alpha systems; for the DECstation,
the limit is three (and you have an additional one onboard).

 Additionally, there exist dual-channel SCSI cards, namely PMAZB-A and
PMAZC-A, of which the former uses NCR 53C84 chips and the latter one uses
NCR 53C84F ones which are handled by the NCR53C9x.c driver.  So
theoretically up to 14 HBAs of this type may exist in a single system (2
are onboard on most TC Alphas). ;-)

  Maciej

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Houten K.H.C. van (Karel)]

Hi Maciej,

"Maciej W. Rozycki" writes:
>
> OK, more writeback fixes.  Please get the following patches: 
>
>- patch-mips-2.4.18-20020530-mb-wb-8.gz,
>
>- patch-mips-2.4.18-20020625-wbflush-7.gz
>
>from 'ftp://ftp.ds2.pg.gda.pl/pub/macro/linux/' and replace the hack I
>sent you yesterday with the following real fix.  After applying the three
>patches you need to rebuild the kernel from scratch, i.e. do `make
>oldconfig dep clean boot modules' as the two above patches modify the
>kernel's configuration.
>
> Please report if this works or not. 

Yes! Now I can use the TC PMAZ-AA without problems. I've copied
some data around, without any problems. Fsck found no problems
on the fs on that drive.

Thanks a lot.

Will these patches go into the oss CVS (2.4) tree?

BTW: Are your patches to the declance driver in a way that we can use
one unified driver for the /240 and the /200 systems? I still use
the driver modified by Dave for my /200 kernels.

Regards,
Karel.

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Maciej W. Rozycki]

Hello Karel,

> > OK, more writeback fixes.  Please get the following patches: 
> >
> >- patch-mips-2.4.18-20020530-mb-wb-8.gz,
> >
> >- patch-mips-2.4.18-20020625-wbflush-7.gz
> >
> >from 'ftp://ftp.ds2.pg.gda.pl/pub/macro/linux/' and replace the hack I
> >sent you yesterday with the following real fix.  After applying the three
> >patches you need to rebuild the kernel from scratch, i.e. do `make
> >oldconfig dep clean boot modules' as the two above patches modify the
> >kernel's configuration.
> >
> > Please report if this works or not. 
> 
> Yes! Now I can use the TC PMAZ-AA without problems. I've copied
> some data around, without any problems. Fsck found no problems
> on the fs on that drive.

 Excellent.  I worried data corruption will happen sooner or later without
these patches, OTOH -- I'm using them since early February, when the new
IRQ code let me discover a deficiency of our synchronization primitives. 

> Will these patches go into the oss CVS (2.4) tree?

 I hope so.  I'm begging Ralf for about half a year, sigh...  The subject
was beaten to death at the list and the feedback looked positive to me
(after a few doubts were resolved), but Ralf seems to be unhappy with the
changes due to some ia64 interactions (I'm still not sure which ones,
though).  Ralf, could you please elaborate? 

 Otherwise, your case convinces me I should not care about purity or
cross-platform consistency of code in this area, anymore.  I've been
observing problems with interrupts due to the lack of iomem access
synchronization already, but unlike for your PMAZ-A problem, their result
was more of annoyance than instability.  Since stability is a priority,
although reluctantly, I will rework the changes to apply to the DECstation
code only, to keep others happy.  The interface won't change, apart from
resolving namespace clashes.

> BTW: Are your patches to the declance driver in a way that we can use
> one unified driver for the /240 and the /200 systems? I still use
> the driver modified by Dave for my /200 kernels.

 Nope, sorry.  Either we miss some hardware configuration data or DEC did
the worst mess possible with the wiring of the LANCE chip (additionally
cropping 8 high bits of the LANCE address space which forces us to waste
128kB of memory for a static, contiguous buffer, instead of using the zone
allocator).  With the current setup it's tough to get both configurations
supported with a single driver (hmm, I wonder how they solved it in
Ultrix...) and I'm not sure it's worth the hassle just for 2.4.  For 2.5
I'll try to separate drivers' frontends to work with the common 7990 core. 
Since each of the PMAX/PMIN, PMAD-A and I/O ASIC setups uses a completely
different chip wiring, there will likely be three separate frontends.  I
may backport the changes to 2.4 eventually if there is still any interest
at that stage.

 I have a patch to sync the Dave's driver to the mainstream declance.c. 
The patch became a bit outdated (not much, though), but I may update it if
there is any interest.  I've sent it to Dave, but not to the list.  I'm
not sure if he applied it.  I have made it available at: 
'ftp://ftp.ds2.pg.gda.pl/pub/macro/linux/patch-mips-2.4.18-20020412-declance-pmad-11.gz'. 

 Politeness keeps me from stating publicly what I think of the designer of
the network hardware in DECstations. 

  Maciej

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Ralf Baechle]

On Mon, Jul 22, 2002 at 02:21:13PM +0200, Maciej W. Rozycki wrote:

>  I hope so.  I'm begging Ralf for about half a year, sigh...  The subject
> was beaten to death at the list and the feedback looked positive to me
> (after a few doubts were resolved), but Ralf seems to be unhappy with the
> changes due to some ia64 interactions (I'm still not sure which ones,
> though).  Ralf, could you please elaborate? 
> 
>  Otherwise, your case convinces me I should not care about purity or
> cross-platform consistency of code in this area, anymore.  I've been
> observing problems with interrupts due to the lack of iomem access
> synchronization already, but unlike for your PMAZ-A problem, their result
> was more of annoyance than instability.  Since stability is a priority,
> although reluctantly, I will rework the changes to apply to the DECstation
> code only, to keep others happy.  The interface won't change, apart from
> resolving namespace clashes.

We had some discussion with the IA64 guys at SGI on how to handle this
kind of I/O ordering issues.  We never came to a final conclusion but
the proposal was the introduction of separate memory barriers macros for
I/O stuff.  Anyway, I think for the moment we should go with your proposal.

  Ralf

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Maciej W. Rozycki]

On Mon, 22 Jul 2002, Ralf Baechle wrote:

> We had some discussion with the IA64 guys at SGI on how to handle this
> kind of I/O ordering issues.  We never came to a final conclusion but
> the proposal was the introduction of separate memory barriers macros for
> I/O stuff.  Anyway, I think for the moment we should go with your proposal.

 Well, nothing stops code from evolving -- if there is a need for a more
finegrained choice of macros, we may fulfill it, don't we?

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

Re: DECStation: Support for PMAZ-AA TC SCSI card?

[Ralf Baechle]

On Mon, Jul 22, 2002 at 07:32:04PM +0200, Maciej W. Rozycki wrote:

> > We had some discussion with the IA64 guys at SGI on how to handle this
> > kind of I/O ordering issues.  We never came to a final conclusion but
> > the proposal was the introduction of separate memory barriers macros for
> > I/O stuff.  Anyway, I think for the moment we should go with your proposal.
> 
>  Well, nothing stops code from evolving -- if there is a need for a more
> finegrained choice of macros, we may fulfill it, don't we?

Frequently a 80% solution is the worst possible because the remaining
unsolved 20% are not enough inclination to get things fixed.  However in
this particular case I agree with you.

  Ralf