Skip to contentSkip to conversation

Oddities with CVS Kernels, Memory on Indigo2

DISCUSSION Kumba 3 participants 12 messages Apr 13, 2003 – Apr 15, 2003 MIPS

lore
Share & tools
lore

Kumba

author
#0 lore raw

Greetings,

Couple issues I've tried to bring up on this ML, but I think was
blocked due my having an AOL.com address (which most people on this list probably categorize as spam). Anyways...

The two issues I've run into are my failure to get a kernel compiled
from linux-mips.org CVS to actually boot, and linux, when I use a working kernel, does not detect the full 320MB of ram in my machine.

On the first issue, the kernel builds fine, just does not boot. There
is no output from the kernel on why it refuses to boot. It just stops responding to any input, and I hear no disk activity from the machine indicating it's actually doing something. I even tried with with an antiquitated 2.4.3 kernel apparently off CVS as well, same result. However, using a stock kernel off kernel.org plus a debian patch for 2.4.19, a workable kernel can be built and it boots fine. I've since modified this debian patch to work with newer kernels (up to 2.4.21-pre7), and it works, no problems. I have no further idea what is wrong in this case.

On the second issue, a "hinv" command at the PROM Monitor lists my
machine as having 320MB of ram, however the kernel is apparently only detecting 256. Upon the advice of Keith Wesolowski of Project Foobazco, I enabled the DEBUG #define in arch/mip/arc/memory.c to see what the ARC Memory descriptor dump was like. Keith said the dump indicated that whatever was going on between the kernel and the PROM/ARC, the kernel was specifically being told there is 256MB of ram.

For reference, the first eight slots are filled with 32meg SIMMs, and
the last four filled with four 16MB simms, for a total of 320MB. I tried testing only the four 16MB simms in the first bank, and both hinv and the kernel reported a functional amount of ram roughly equal to 64MB. I tried putting 128MB in bank 1, 64MB in bank 2, and the remaining 128MB in bank 3, thinking linux just doesn't see the last bank of ram...ARC/PROM still reported 320MB, the kernel said 256MB.

Really I'm at a loss as to what is wrong. I do know that when I
briefly had installed an IP28 mainboard + R10K module, and all the ram and booted Thiemo's experimental 2.5.1 IP28 kernel, it saw all 320MB of ram. However, that being a totally different mainboard, such information is probably not pertinent to this little investigation here...

I've included a log of what hinv says, my attempt to boot a CVS kernel,
and a working kernel. Of notice, is with the CVS kernel, I also enabled DEBUG in arch/mips/arc/memory.c, and that did print debug output , however nothing further to indicate the CVS kernel booting. My PROM chip also says this information: P/N: 070-1367-010, 3895 S6275. I include that only on the odd circumstance I have an extremely weird PROM. Who knows...

If any other information is needed, let me know, I'll provide whatever
I can. It may not seem like much, it's only another 64MB of ram not available, but it's definitely out of the norm, so I figure it's worth a good look into.

--Kumba
Gentoo/Sparc/Mips Developer

// hinv output

System: IP22
Processor: 250 Mhz R4400, with FPU
Primary I-cache size: 16 Kbytes
Primary D-cache size: 16 Kbytes
Secondary cache size: 2048 Kbytes
Memory size: 320 Mbytes
SCSI Disk: scsi(0)disk(1)
SCSI Disk: scsi(0)disk(2)
SCSI CDROM: scsi(0)cdrom(4)
Audio: Iris Audio Processor: version A2 revision 1.1.0

// linux-mips CVS kernel Boot attempt

Quoted 1 line
Quoted 1 line

boot -f 2421pm

ARCS MEMORY DESCRIPTOR dump:
[0,a8748a48]: base<00000000> pages<00000001> type<Exception Block> [1,a8748a64]: base<00000001> pages<00000001> type<ARCS Romvec Page> [2,a8748a2c]: base<00008002> pages<000001dd> type<Standalone Program Pages> [3,a87492fc]: base<000081df> pages<00000561> type<Generic Free RAM> [4,a87492cc]: base<00008740> pages<000000c0> type<ARCS Temp Storage Area> [5,a87492b0]: base<00008800> pages<0000f800> type<Generic Free RAM>

// 2.4.21-pre7 off kernel.org + modified debian patch

Quoted 1 line
Quoted 1 line

boot -f 2421p7d

ARCS MEMORY DESCRIPTOR dump:
[0,a8748a48]: base<00000000> pages<00000001> type<Exception Block> [1,a8748a64]: base<00000001> pages<00000001> type<ARCS Romvec Page> [2,a8748a2c]: base<00008002> pages<000001ee> type<Standalone Program Pages> [3,a87492fc]: base<000081f0> pages<00000550> type<Generic Free RAM> [4,a87492cc]: base<00008740> pages<000000c0> type<ARCS Temp Storage Area> [5,a87492b0]: base<00008800> pages<0000f800> type<Generic Free RAM>
ARCH: SGI-IP22
PROMLIB: ARC firmware Version 1 Revision 10
CPU: MIPS-R4400 FPU<MIPS-R4400FPC> ICACHE DCACHE SCACHE
CPU revision is: 00000460
FPU revision is: 00000500
Primary instruction cache 16kb, linesize 16 bytes.
Primary data cache 16kb, linesize 16 bytes.
Secondary cache sized at 2048K linesize 128 bytes.
Linux version 2.4.21-pre7 (root@isengard) (gcc version 3.2.2) #2 Sat Apr 12 17:3 0:18 EDT 2003
MC: SGI memory controller Revision 3
Determined physical RAM map:
memory: 00001000 @ 00000000 (reserved)
memory: 00001000 @ 00001000 (reserved)
memory: 001ee000 @ 08002000 (reserved)
memory: 00550000 @ 081f0000 (usable)
memory: 000c0000 @ 08740000 (ROM data)
memory: 0f800000 @ 08800000 (usable)
On node 0 totalpages: 98304
zone(0): 36864 pages.
zone(1): 61440 pages.
zone(2): 0 pages.
Kernel command line: root=/dev/sda3
EISA: Probing bus...
EISA: slot 4 : TCM5970 detected.
EISA: Detected 1 card.
ISA support compiled in.
Calibrating system timer... 1250000 [250.00 MHz CPU]
GIO: Scanning for GIO cards...
GIO: Card 0x7f @ 0x1f000000
GIO: Card 0x04 @ 0x1f400000
Console: colour dummy device 80x25
zs0: console input
Console: ttyS0 (Zilog8530), 38400 baud
Calibrating delay loop... 124.92 BogoMIPS
Memory: 255108k/259392k available (1603k kernel code, 4284k reserved, 108k data,
88k init, 0k highmem)
Dentry cache hash table entries: 65536 (order: 7, 524288 bytes) Inode cache hash table entries: 32768 (order: 6, 262144 bytes) Mount cache hash table entries: 512 (order: 0, 4096 bytes) Buffer-cache hash table entries: 32768 (order: 5, 131072 bytes) Page-cache hash table entries: 131072 (order: 7, 524288 bytes) Checking for 'wait' instruction... unavailable.
POSIX conformance testing by UNIFIX
isapnp: Scanning for PnP cards...
isapnp: No Plug & Play device found
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: v1.12c (20020818) Richard Gooch ([email protected])
devfs: boot_options: 0x1
parport0: PC-style at 0x278 [PCSPP,TRISTATE,EPP]
initialize_kbd: Keyboard failed self test
keyboard: Timeout - AT keyboard not present?(ed)
keyboard: Timeout - AT keyboard not present?(f4)
pty: 256 Unix98 ptys configured
DS1286 Real Time Clock Driver v1.0
lp0: using parport0 (polling).
tipar: parallel link cable driver, version 1.18
tipar: registering to devfs : major = 115, minor = 0, node = 0
tipar0: using parport0 (polling).
tipar0: link cable not found.
Hardware Watchdog Timer for SGI IP22: 0.2
sgiseeq.c: David S. Miller ([email protected])
eth0: SGI Seeq8003 08:00:69:0a:6d:33
RAMDISK driver initialized: 16 RAM disks of 4096K size 1024 blocksize
loop: loaded (max 8 devices)
3c59x: Donald Becker and others. www.scyld.com/network/vortex.html See Documentation/networking/vortex.txt
3c59x: 3Com EISA 3c590 Vortex 10Mbps at 0x4000. Vers LK1.1.16
ff:ff:ff:ff:ff:ff, IRQ 3
product code ffff rev ffff.15 date 15-31-127
Full duplex capable
1024K word-wide RAM 3:5 Rx:Tx split, autoselect/<invalid transceiver> interfac
e.
Enabling bus-master transmits and early receives.
3c59x: scatter/gather enabled. h/w checksums disabled
SCSI subsystem driver Revision: 1.00
wd33c93-0: chip=WD33c93B/13 no_sync=0xff no_dma=0 debug_flags=0x00

           setup_args=,,,,,,,,,
           Version 1.25 - 09/Jul/1997, Compiled Apr 12 2003 at 16:25:26
wd33c93-1: chip=WD33c93B/13 no_sync=0xff no_dma=0 debug_flags=0x00
           setup_args=,,,,,,,,,
           Version 1.25 - 09/Jul/1997, Compiled Apr 12 2003 at 16:25:26

scsi0 : SGI WD93
scsi1 : SGI WD93
sending SDTR 0103013f0csync_xfer=2c Vendor: SEAGATE Model: ST19171W
Rev: 2219
Type: Direct-Access ANSI SCSI revision: 02
sending SDTR 0103013f0csync_xfer=2c Vendor: IBM Model: DCHS09F CLAR09
Rev: SG53
Type: Direct-Access ANSI SCSI revision: 02
sending SDTR 0103013f08sync_xfer=28 Vendor: TOSHIBA Model: CD-ROM XM-5701TA
Rev: 1557
Type: CD-ROM ANSI SCSI revision: 02
Attached scsi disk sda at scsi0, channel 0, id 1, lun 0
Attached scsi disk sdb at scsi0, channel 0, id 2, lun 0
SCSI device sda: 17783112 512-byte hdwr sectors (9105 MB)
Partition check:
/dev/scsi/host0/bus0/target1/lun0: p1 p2 p3 p4 p5 p6 p7 p8 p9
SCSI device sdb: 17796078 512-byte hdwr sectors (9112 MB)
/dev/scsi/host0/bus0/target2/lun0: p1 p2 p3 p4
Attached scsi CD-ROM sr0 at scsi0, channel 0, id 4, lun 0
scsi0 channel 0 : resetting for second half of retries.
SCSI bus is being reset for host 0 channel 0.
scsi0: reset. sending SDTR 0103013f08sync_xfer=28<3>sr0: CDROM (ioctl) error, c
ommand: 0x1a 00 2a 00 80 00
sr00:00: sns = 0 0
Non-extended sense class 0 code 0x0
Raw sense data:0x00 0x00 0x00 0x00
sr0: scsi-1 drive
Uniform CD-ROM driver Revision: 3.12
SGI Zilog8530 serial driver version 1.00
tty00 at 0xbfbd9830 (irq = 45) is a Zilog8530
tty01 at 0xbfbd9838 (irq = 45) is a Zilog8530
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP
IP: routing cache hash table of 4096 buckets, 32Kbytes
TCP: Hash tables configured (established 32768 bind 65536)
NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
sending SDTR 0103013f0csync_xfer=2c<6>kjournald starting. Commit interval 5 se
conds
EXT3-fs: mounted filesystem with ordered data mode.
VFS: Mounted root (ext3 filesystem) readonly.
Mounted devfs on /dev
Freeing prom memory: 768kb freed
INIT: version 2.84 booting

Gentoo Linux; http://www.gentoo.org/
Copyright 2001-2003 Gentoo Technologies, Inc.; Distributed under the GPL

[snip]

#1 lore raw
Quoted 1 line · Kumba

Greetings,

Hi,

Quoted 8 lines

Couple issues I've tried to bring up on this ML, but I think was
blocked due my having an AOL.com address (which most people on this list probably categorize as spam). Anyways...

The two issues I've run into are my failure to get a kernel compiled
from linux-mips.org CVS to actually boot, and linux, when I use a working kernel, does not detect the full 320MB of ram in my machine.

Quoted 10 lines

On the first issue, the kernel builds fine, just does not boot. There is no output from the kernel on why it refuses to boot. It just
stops responding to any input, and I hear no disk activity from the machine indicating it's actually doing something. I even tried with with an antiquitated 2.4.3 kernel apparently off CVS as well, same result. However, using a stock kernel off kernel.org plus a debian patch for 2.4.19, a workable kernel can be built and it boots fine. I've since modified this debian patch to work with newer kernels (up to 2.4.21-pre7), and it works, no problems. I have no further idea what is wrong in this case.

I'd say you've tried cvs kernel at the times when support for R4400 caches was broken. I put kernel I'm currently running at
http://www.linux-mips.org/~ladis/vmlinux.gz (gunzip it :)), as you can see from dmesg at the end of this mail CPU used in your machine is the same. If kernel I provided doesn't boot for you I'd like to ask you for help with debugging (kernel was build from cvs updated at 8:30 CEST)

Quoted 7 lines

On the second issue, a "hinv" command at the PROM Monitor lists my
machine as having 320MB of ram, however the kernel is apparently only detecting 256. Upon the advice of Keith Wesolowski of Project Foobazco, I enabled the DEBUG #define in arch/mip/arc/memory.c to see what the ARC Memory descriptor dump was like. Keith said the dump indicated that whatever was going on between the kernel and the PROM/ARC, the kernel was specifically being told there is 256MB of ram.

This is known bug, but unfortunately I have not enough RAM to meet it...

Quoted 116 lines

For reference, the first eight slots are filled with 32meg SIMMs, and the last four filled with four 16MB simms, for a total of 320MB. I
tried testing only the four 16MB simms in the first bank, and both hinv and the kernel reported a functional amount of ram roughly equal to 64MB. I tried putting 128MB in bank 1, 64MB in bank 2, and the remaining 128MB in bank 3, thinking linux just doesn't see the last bank of ram...ARC/PROM still reported 320MB, the kernel said 256MB.

Really I'm at a loss as to what is wrong. I do know that when I
briefly had installed an IP28 mainboard + R10K module, and all the ram and booted Thiemo's experimental 2.5.1 IP28 kernel, it saw all 320MB of ram. However, that being a totally different mainboard, such information is probably not pertinent to this little investigation here...

I've included a log of what hinv says, my attempt to boot a CVS kernel, and a working kernel. Of notice, is with the CVS kernel, I also
enabled DEBUG in arch/mips/arc/memory.c, and that did print debug output , however nothing further to indicate the CVS kernel booting. My PROM chip also says this information: P/N: 070-1367-010, 3895 S6275. I include that only on the odd circumstance I have an extremely weird PROM. Who knows...

If any other information is needed, let me know, I'll provide whatever I can. It may not seem like much, it's only another 64MB of ram
not available, but it's definitely out of the norm, so I figure it's worth a good look into.

--Kumba
Gentoo/Sparc/Mips Developer

// hinv output

System: IP22
Processor: 250 Mhz R4400, with FPU
Primary I-cache size: 16 Kbytes
Primary D-cache size: 16 Kbytes
Secondary cache size: 2048 Kbytes
Memory size: 320 Mbytes
SCSI Disk: scsi(0)disk(1)
SCSI Disk: scsi(0)disk(2)
SCSI CDROM: scsi(0)cdrom(4)
Audio: Iris Audio Processor: version A2 revision 1.1.0

// linux-mips CVS kernel Boot attempt

Quoted 1 line
Quoted 1 line

boot -f 2421pm

ARCS MEMORY DESCRIPTOR dump:
[0,a8748a48]: base<00000000> pages<00000001> type<Exception Block> [1,a8748a64]: base<00000001> pages<00000001> type<ARCS Romvec Page> [2,a8748a2c]: base<00008002> pages<000001dd> type<Standalone Program Pages> [3,a87492fc]: base<000081df> pages<00000561> type<Generic Free RAM> [4,a87492cc]: base<00008740> pages<000000c0> type<ARCS Temp Storage Area> [5,a87492b0]: base<00008800> pages<0000f800> type<Generic Free RAM>

// 2.4.21-pre7 off kernel.org + modified debian patch

Quoted 1 line
Quoted 1 line

boot -f 2421p7d

ARCS MEMORY DESCRIPTOR dump:
[0,a8748a48]: base<00000000> pages<00000001> type<Exception Block> [1,a8748a64]: base<00000001> pages<00000001> type<ARCS Romvec Page> [2,a8748a2c]: base<00008002> pages<000001ee> type<Standalone Program Pages> [3,a87492fc]: base<000081f0> pages<00000550> type<Generic Free RAM> [4,a87492cc]: base<00008740> pages<000000c0> type<ARCS Temp Storage Area> [5,a87492b0]: base<00008800> pages<0000f800> type<Generic Free RAM>
ARCH: SGI-IP22
PROMLIB: ARC firmware Version 1 Revision 10
CPU: MIPS-R4400 FPU<MIPS-R4400FPC> ICACHE DCACHE SCACHE
CPU revision is: 00000460
FPU revision is: 00000500
Primary instruction cache 16kb, linesize 16 bytes.
Primary data cache 16kb, linesize 16 bytes.
Secondary cache sized at 2048K linesize 128 bytes.
Linux version 2.4.21-pre7 (root@isengard) (gcc version 3.2.2) #2 Sat Apr 12 17:3 0:18 EDT 2003
MC: SGI memory controller Revision 3
Determined physical RAM map:
memory: 00001000 @ 00000000 (reserved)
memory: 00001000 @ 00001000 (reserved)
memory: 001ee000 @ 08002000 (reserved)
memory: 00550000 @ 081f0000 (usable)
memory: 000c0000 @ 08740000 (ROM data)
memory: 0f800000 @ 08800000 (usable)
On node 0 totalpages: 98304
zone(0): 36864 pages.
zone(1): 61440 pages.
zone(2): 0 pages.
Kernel command line: root=/dev/sda3
EISA: Probing bus...
EISA: slot 4 : TCM5970 detected.
EISA: Detected 1 card.
ISA support compiled in.
Calibrating system timer... 1250000 [250.00 MHz CPU]
GIO: Scanning for GIO cards...
GIO: Card 0x7f @ 0x1f000000
GIO: Card 0x04 @ 0x1f400000
Console: colour dummy device 80x25
zs0: console input
Console: ttyS0 (Zilog8530), 38400 baud
Calibrating delay loop... 124.92 BogoMIPS
Memory: 255108k/259392k available (1603k kernel code, 4284k reserved, 108k data,
88k init, 0k highmem)
Dentry cache hash table entries: 65536 (order: 7, 524288 bytes) Inode cache hash table entries: 32768 (order: 6, 262144 bytes) Mount cache hash table entries: 512 (order: 0, 4096 bytes) Buffer-cache hash table entries: 32768 (order: 5, 131072 bytes) Page-cache hash table entries: 131072 (order: 7, 524288 bytes) Checking for 'wait' instruction... unavailable.
POSIX conformance testing by UNIFIX
isapnp: Scanning for PnP cards...
isapnp: No Plug & Play device found

Quoted 8 lines

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: v1.12c (20020818) Richard Gooch ([email protected])
devfs: boot_options: 0x1
parport0: PC-style at 0x278 [PCSPP,TRISTATE,EPP]

            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

it can't work ;)

Quoted 67 lines

initialize_kbd: Keyboard failed self test
keyboard: Timeout - AT keyboard not present?(ed)
keyboard: Timeout - AT keyboard not present?(f4)
pty: 256 Unix98 ptys configured
DS1286 Real Time Clock Driver v1.0
lp0: using parport0 (polling).
tipar: parallel link cable driver, version 1.18
tipar: registering to devfs : major = 115, minor = 0, node = 0
tipar0: using parport0 (polling).
tipar0: link cable not found.
Hardware Watchdog Timer for SGI IP22: 0.2
sgiseeq.c: David S. Miller ([email protected])
eth0: SGI Seeq8003 08:00:69:0a:6d:33
RAMDISK driver initialized: 16 RAM disks of 4096K size 1024 blocksize
loop: loaded (max 8 devices)
3c59x: Donald Becker and others. www.scyld.com/network/vortex.html See Documentation/networking/vortex.txt
3c59x: 3Com EISA 3c590 Vortex 10Mbps at 0x4000. Vers LK1.1.16
ff:ff:ff:ff:ff:ff, IRQ 3
product code ffff rev ffff.15 date 15-31-127
Full duplex capable
1024K word-wide RAM 3:5 Rx:Tx split, autoselect/<invalid transceiver> interfac
e.
Enabling bus-master transmits and early receives.
3c59x: scatter/gather enabled. h/w checksums disabled
SCSI subsystem driver Revision: 1.00
wd33c93-0: chip=WD33c93B/13 no_sync=0xff no_dma=0 debug_flags=0x00

           setup_args=,,,,,,,,,
           Version 1.25 - 09/Jul/1997, Compiled Apr 12 2003 at 16:25:26
wd33c93-1: chip=WD33c93B/13 no_sync=0xff no_dma=0 debug_flags=0x00
           setup_args=,,,,,,,,,
           Version 1.25 - 09/Jul/1997, Compiled Apr 12 2003 at 16:25:26

scsi0 : SGI WD93
scsi1 : SGI WD93
sending SDTR 0103013f0csync_xfer=2c Vendor: SEAGATE Model: ST19171W
Rev: 2219
Type: Direct-Access ANSI SCSI revision: 02
sending SDTR 0103013f0csync_xfer=2c Vendor: IBM Model: DCHS09F CLAR09
Rev: SG53
Type: Direct-Access ANSI SCSI revision: 02
sending SDTR 0103013f08sync_xfer=28 Vendor: TOSHIBA Model: CD-ROM XM-5701TA
Rev: 1557
Type: CD-ROM ANSI SCSI revision: 02
Attached scsi disk sda at scsi0, channel 0, id 1, lun 0
Attached scsi disk sdb at scsi0, channel 0, id 2, lun 0
SCSI device sda: 17783112 512-byte hdwr sectors (9105 MB)
Partition check:
/dev/scsi/host0/bus0/target1/lun0: p1 p2 p3 p4 p5 p6 p7 p8 p9
SCSI device sdb: 17796078 512-byte hdwr sectors (9112 MB)
/dev/scsi/host0/bus0/target2/lun0: p1 p2 p3 p4
Attached scsi CD-ROM sr0 at scsi0, channel 0, id 4, lun 0
scsi0 channel 0 : resetting for second half of retries.
SCSI bus is being reset for host 0 channel 0.
scsi0: reset. sending SDTR 0103013f08sync_xfer=28<3>sr0: CDROM (ioctl) error, c
ommand: 0x1a 00 2a 00 80 00
sr00:00: sns = 0 0
Non-extended sense class 0 code 0x0
Raw sense data:0x00 0x00 0x00 0x00
sr0: scsi-1 drive
Uniform CD-ROM driver Revision: 3.12
SGI Zilog8530 serial driver version 1.00
tty00 at 0xbfbd9830 (irq = 45) is a Zilog8530
tty01 at 0xbfbd9838 (irq = 45) is a Zilog8530
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP
IP: routing cache hash table of 4096 buckets, 32Kbytes
TCP: Hash tables configured (established 32768 bind 65536)

Quoted 8 lines

NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
sending SDTR 0103013f0csync_xfer=2c<6>kjournald starting. Commit interval 5 se
conds
EXT3-fs: mounted filesystem with ordered data mode.
VFS: Mounted root (ext3 filesystem) readonly.
Mounted devfs on /dev
Freeing prom memory: 768kb freed
INIT: version 2.84 booting

[snip]

ARCH: SGI-IP22
PROMLIB: ARC firmware Version 1 Revision 10
CPU: MIPS-R4400 FPU<MIPS-R4400FPC> ICACHE DCACHE SCACHE
CPU revision is: 00000460
FPU revision is: 00000500
Primary instruction cache 16kb, physically tagged, direct mapped, linesize 16 bytes Primary data cache 16kb direct mapped, linesize 16 bytes
Secondary cache sized at 1024K, linesize 128 bytes.
Linux version 2.4.21-pre4 (ladis@simek) (gcc version 3.2) #11 Po dub 14 09:10:06 CEST 2003
MC: SGI memory controller Revision 3
Determined physical RAM map:
memory: 00001000 @ 00000000 (reserved)
memory: 00001000 @ 00001000 (reserved)
memory: 001be000 @ 08002000 (reserved)
memory: 00580000 @ 081c0000 (usable)
memory: 000c0000 @ 08740000 (ROM data)
memory: 03800000 @ 08800000 (usable)
On node 0 totalpages: 49152
zone(0): 49152 pages.
zone(1): 0 pages.
zone(2): 0 pages.
Kernel command line: ip=any
Calibrating system timer... 1000000 [200.00 MHz CPU]
Console: colour dummy device 80x25
zs0: console input
Console: ttyS0 (Zilog8530), 9600 baud
Calibrating delay loop... 99.94 BogoMIPS
Memory: 60816k/62976k available (1431k kernel code, 2160k reserved, 104k 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: 512 (order: 0, 4096 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
Linux NET4.0 for Linux 2.4
Based upon Swansea University Computer Society NET3.039
Initializing RT netlink socket
Starting kswapd
initialize_kbd: Keyboard failed self test
pty: 256 Unix98 ptys configured
keyboard: Timeout - AT keyboard not present?(ed)
keyboard: Timeout - AT keyboard not present?(f4)
DS1286 Real Time Clock Driver v1.0
SGI Zilog8530 serial driver version 1.00
tty00 at 0xbfbd9830 (irq = 45) is a Zilog8530
tty01 at 0xbfbd9838 (irq = 45) is a Zilog8530
sgiseeq.c: David S. Miller ([email protected])
eth0: SGI Seeq8003 08:00:69:08:ad:02
SCSI subsystem driver Revision: 1.00
wd33c93-0: chip=WD33c93B/13 no_sync=0xff no_dma=0 debug_flags=0x00

           setup_args=,,,,,,,,,
           Version 1.25 - 09/Jul/1997, Compiled Apr 14 2003 at 09:11:31
wd33c93-1: chip=WD33c93B/13 no_sync=0xff no_dma=0 debug_flags=0x00
           setup_args=,,,,,,,,,
           Version 1.25 - 09/Jul/1997, Compiled Apr 14 2003 at 09:11:31

scsi0 : SGI WD93
scsi1 : SGI WD93
SGI HAL2 revision 1.1.0
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 32768) Sending BOOTP requests . OK
IP-Config: Got BOOTP answer from 192.168.50.22, my address is 192.168.50.56
IP-Config: Complete:

      device=eth0, addr=192.168.50.56, mask=255.255.255.0, gw=192.168.50.1,
     host=indigo2, domain=ckdenergo.cz, nis-domain=(none),
     bootserver=192.168.50.22, rootserver=192.168.50.22, rootpath=/exports/ip22

NET4: Unix domain sockets 1.0/SMP for Linux NET4.0. Looking up port of RPC 100003/2 on 192.168.50.22 Looking up port of RPC 100005/1 on 192.168.50.22
VFS: Mounted root (nfs filesystem).
Freeing prom memory: 768kb freed
Freeing unused kernel memory: 76k freed

regards,
ladis

author
#2 lore raw
Quoted 7 lines · Ladislav Michl

I'd say you've tried cvs kernel at the times when support for R4400 caches was broken. I put kernel I'm currently running at
http://www.linux-mips.org/~ladis/vmlinux.gz (gunzip it :)), as you can see from dmesg at the end of this mail CPU used in your machine is the same. If kernel I provided doesn't boot for you I'd like to ask you for help with debugging (kernel was build from cvs updated at 8:30 CEST)

Wow, this booted. Several people I talked to thought it was originally a serial console issue. Judging by the several times I've chosen to build a kernel, it seems R4400 cache gets broken quite often. I'll run a cvs sync now and try to build my own kernel, since this appears to be built from recent code. Your kernel lacked a few things gentoo requires to boot, but it at least proves I'm not going insane over here.

Quoted 1 line

This is known bug, but unfortunately I have not enough RAM to meet it...

A known bug? Interesting. I mentioned it many moons ago in #mipslinux, and Ralf seemed curious about it, but he said he looked in the kernel where the memory was detected and said he didn't see anything wrong there. Anything I can do to help look into this bug and possibly fix?

Quoted 4 lines
Quoted 1 line

parport0: PC-style at 0x278 [PCSPP,TRISTATE,EPP]

            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

it can't work ;)

It worked fine for me with printing :)

[root@angband root]# echo "Hello World" > /dev/lp0 [root@angband root]# echo "Wow, It works" > /dev/lp0 [root@angband root]# echo "Totally Cool" > /dev/lp0

^--- All printed out on paper, although scratchy. I need new ink cartridges. Canon BJC-620

Mind you, that's an ISA Parallel Port card I dropped in. I noticed the SGI's parallel port never worked, so I dug up a spare and tried it.

--Kumba

#3 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"kumba" == kumba <[email protected]> writes:

kumba> Mind you, that's an ISA Parallel Port card I dropped in. I kumba> noticed the SGI's parallel port never worked, so I dug up a kumba> spare and tried it.

So you're the first to try an ISA card on the I2. I must say I'm quite pleased it worked ! :-)

        M.
Signature · 1 line
Places change, faces change. Life is so very strange.
author
#4 lore raw

As am I. I've also gotten an NE2000 ISA 10mbps network card to be
detected and work under `ifconfig', but forgot how to deal with multiple network cards, so I didn't actually get it hooked up to the network. I've got a 3com 3c597 EISA card in there at the moment, but I think it's cooked, since it's MAC Address reports itself as all ff's. I'm also hunting for an EISA Mach32 video card to see if maybe on the offchance, it's possible to build a VESA Compatible framebuffer for the system. That will prove to be an interesting experiment.

--Kumba

Quoted 11 lines · Marc Zyngier
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"kumba" == kumba <[email protected]> writes:

kumba> Mind you, that's an ISA Parallel Port card I dropped in. I kumba> noticed the SGI's parallel port never worked, so I dug up a kumba> spare and tried it.

So you're the first to try an ISA card on the I2. I must say I'm quite pleased it worked ! :-)

        M.
author
#6 lore raw
In reply to Marc Zyngier

Also, forgot to mention on this topic, but while messing with ISA/EISA
cards in the I2, I've run across some strange "hack" regarding Local IRQ 3 on the machine. There's a construct inside arch/mips/sgi-ip22/ip22-int.c in the enable_local3_irq() function that purposely panics the kernel if LIRQ3 is probed or used. Any one got any idea why this is? There aren't any comments in the code to explain this odd little construct, and removing it generates some amusing messages at bootup, long the lines of "Whee: Got an LIO3 irq, winging it...". Quite odd if you ask me.

--Kumba

Quoted 11 lines · Marc Zyngier
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"kumba" == kumba <[email protected]> writes:

kumba> Mind you, that's an ISA Parallel Port card I dropped in. I kumba> noticed the SGI's parallel port never worked, so I dug up a kumba> spare and tried it.

So you're the first to try an ISA card on the I2. I must say I'm quite pleased it worked ! :-)

        M.
#5 lore raw
In reply to Kumba
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"kumba" == kumba <[email protected]> writes:

kumba> As am I. I've also gotten an NE2000 ISA 10mbps network card to kumba> be detected and work under `ifconfig', but forgot how to deal kumba> with multiple network cards, so I didn't actually get it hooked kumba> up to the network.

Very nice, indeed.

kumba> I've got a 3com 3c597 EISA card in there at the moment, but I kumba> think it's cooked, since it's MAC Address reports itself as all kumba> ff's.

Maybe not. 3c597 may be tricky to support, because it does bus-mastering, which is really a no-go given the state of the current EISA code on IP22. I'll try to do something about it when 2.5 is running on the IP22 (if it ever runs...).

kumba> I'm also hunting for an EISA Mach32 video card to see if maybe kumba> on the offchance, it's possible to build a VESA Compatible kumba> framebuffer for the system. That will prove to be an kumba> interesting experiment.

Interesting is quite an understatement... :-)

        M.
Signature · 1 line
Places change, faces change. Life is so very strange.
#7 lore raw
In reply to Kumba
Quoted 10 lines · Kumba

Also, forgot to mention on this topic, but while messing with ISA/EISA
cards in the I2, I've run across some strange "hack" regarding Local IRQ 3 on the machine. There's a construct inside arch/mips/sgi-ip22/ip22-int.c in the enable_local3_irq() function that purposely panics the kernel if LIRQ3 is probed or used. Any one got any idea why this is? There aren't any comments in the code to explain this odd little construct, and removing it generates some amusing messages at bootup, long the lines of "Whee: Got an LIO3 irq, winging it...". Quite odd if you ask me.

Several chips used in Indy (and Indigo2) are used in much complicated machines (not supported by linux) and SGI always designed its machines with modularity in mind. local3_irq is another cascade where nothing is hooked on Indy, so you can't get this irq. and if it happens there is sometning strange with our system. there are no comments because you need to understand it before coding and once you read documentation comments are useless ;-)

ladis

ps. there is driver for built-in parport now by Vincent Stehle
http://vincent.stehle.free.fr/sgi/parport.php3

author
#8 lore raw
Quoted 9 lines · Ladislav Michl

On Mon, Apr 14, 2003 at 12:24:54PM -0400, Kumba wrote:

Several chips used in Indy (and Indigo2) are used in much complicated machines (not supported by linux) and SGI always designed its machines with modularity in mind. local3_irq is another cascade where nothing is hooked on Indy, so you can't get this irq. and if it happens there is sometning strange with our system. there are no comments because you need to understand it before coding and once you read documentation comments are useless ;-)

Quoted 4 lines

ladis

ps. there is driver for built-in parport now by Vincent Stehle
http://vincent.stehle.free.fr/sgi/parport.php3

Ah ha, interesting. I noticed it, because testing both the parallel port and NE2000 ISA card, I had to avoid that IRQ. The parallel port card had physical jumpers I set on it to get around trying to use IRQ3, and the NE2000 Card you *have* to know the IRQ. Can't use the ether=0,0,ethX line on the kernel, because if the kernel attempts to probe for IRQs, it runs into that panic. Enabling it didn't seem to do anything bad either, except that odd message which sounded pretty weird for a kernel.

--Kumba

author
#9 lore raw
In reply to Ladislav Michl

A question with the CVS Kernel. Are there any known issues as to the
specific toolchain used to build it? Currently I'm using GCC 3.2.2 + Glibc 2.3.2 + Binutils 2.13.90.0.16. I tried a CVS updateshortly after testing this kernel, and it still doesn't boot, so either my timing is really bad, or I'm using a toolchain with known issues. I have noticed the toolchain does not support -mcpu in .18 of binutils, and I originally thought this was only going to be a GCC change in 3.3, but my guess if it's the same in .18/.20, and as such I'll need to use the instructions posted about a month about to use -mabi/-march?

--Kumba

Quoted 8 lines · Ladislav Michl

On Sun, Apr 13, 2003 at 01:13:42AM -0400, Kumba wrote:

I'd say you've tried cvs kernel at the times when support for R4400 caches was broken. I put kernel I'm currently running at
http://www.linux-mips.org/~ladis/vmlinux.gz (gunzip it :)), as you can see from dmesg at the end of this mail CPU used in your machine is the same. If kernel I provided doesn't boot for you I'd like to ask you for help with debugging (kernel was build from cvs updated at 8:30 CEST)

#10 lore raw
Quoted 9 lines · Kumba

A question with the CVS Kernel. Are there any known issues as to the
specific toolchain used to build it? Currently I'm using GCC 3.2.2 + Glibc 2.3.2 + Binutils 2.13.90.0.16. I tried a CVS updateshortly after testing this kernel, and it still doesn't boot, so either my timing is really bad, or I'm using a toolchain with known issues. I have noticed the toolchain does not support -mcpu in .18 of binutils, and I originally thought this was only going to be a GCC change in 3.3, but my guess if it's the same in .18/.20, and as such I'll need to use the instructions posted about a month about to use -mabi/-march?

silly question: you specified -r linux_2_4 when checkouting kernel, didn't you? otherwise you are trying to boot 2.5.47, which has broken serial console support.

anyway, toolchains i'm using to build 32-bit kernel:

$ mips-linux-gcc -v
Reading specs from /home/ladis/mips-tools/lib/gcc-lib/mips-linux/3.2/specs
Configured with: ../gcc-3.2/configure --prefix=/home/ladis/mips-tools --disable-shared --with-gnu-as --enable-languages=c --disable-nls --with-newlib --enable-checking=no --disable-threads --with-headers=/home/ladis/src/linux/include --target=mips-linux
Thread model: single
gcc version 3.2
$ mips-linux-ld -v
GNU ld version 2.13

64-bit kernel (egcs is patched with egcs-1.1.2.diff.gz from linux-mips.org ftp):

$ mips64-linux-gcc -v
Reading specs from /home/ladis/mips-tools/lib/gcc-lib/mips64-linux/egcs-2.91.66/specs
gcc version egcs-2.91.66 19990314 (egcs-1.1.2 release)
$ mips64-linux-ld -v
GNU ld version 2.13

ladis

author
#11 lore raw
Quoted 5 lines · Ladislav Michl

On Tue, Apr 15, 2003 at 04:35:23AM -0400, Kumba wrote:

silly question: you specified -r linux_2_4 when checkouting kernel, didn't you? otherwise you are trying to boot 2.5.47, which has broken serial console support.

Yes, I know I'm definitely working with 2.4 sources. 2.5's menuconfig still confuses me somewhat due to it's more organized layout.

Quoted 9 lines

anyway, toolchains i'm using to build 32-bit kernel:

$ mips-linux-gcc -v
Reading specs from /home/ladis/mips-tools/lib/gcc-lib/mips-linux/3.2/specs
Configured with: ../gcc-3.2/configure --prefix=/home/ladis/mips-tools --disable-shared --with-gnu-as --enable-languages=c --disable-nls --with-newlib --enable-checking=no --disable-threads --with-headers=/home/ladis/src/linux/include --target=mips-linux
Thread model: single
gcc version 3.2
$ mips-linux-ld -v
GNU ld version 2.13
Quoted 6 lines

64-bit kernel (egcs is patched with egcs-1.1.2.diff.gz from linux-mips.org ftp):

$ mips64-linux-gcc -v
Reading specs from /home/ladis/mips-tools/lib/gcc-lib/mips64-linux/egcs-2.91.66/specs
gcc version egcs-2.91.66 19990314 (egcs-1.1.2 release)
$ mips64-linux-ld -v
GNU ld version 2.13
# mips-unknown-linux-gnu-gcc -v
Reading specs from /home/crossdev/mips/lib/gcc-lib/mips-unknown-linux-gnu/3.2.2/specs
Configured with: ../configure --prefix=/home/crossdev/mips --target=mips-unknown-linux-gnu --host=sparc-unknown-linux-gnu --disable-multilib --enable-shared --enable-languages=c,c++,ada,f77,objc --enable-nls --without-included-gettext --with-system-zlib --enable-threads=posix --enable-long-long --disable-checking --enable-cstdio=stdio --enable-clocale=generic --enable-__cxa_atexit --enable-version-specific-runtime-libs --with-local-prefix=/local --with-libs=/home/crossdev/mips/lib --with-headers=/home/crossdev/mips/mips-unknown-linux-gnu/include
Thread model: posix
gcc version 3.2.2
# mips-unknown-linux-gnu-ld -v
GNU ld version 2.13.90.0.16 20021126

It's a cross compiler, specifically a sparc -> mips one, as my sparc machine is the fastest machine (Blade 100) running linux on my network atm. I stuck to the .16 binutils initially because I didn't know how to address the -mcpu curiosity that appears in .18 and .20. I originally thought that binutils was broken on mips, but after thinking about it, I partially suspect -mcpu has been gutted from the usable options.

I'm wondering if there are some strange oddities a slightly older binutils such as .16 may have in relation to the CVS kernels. In all likely, it's probably something else I'm overseeing, but I'm trying to eliminate possible causes and get to the root of this problem.

--Kumba