Skip to contentSkip to conversation

could not map ACPI

DISCUSSION Robert Woerle Paceblade/Support 3 participants 8 messages Jan 21, 2003 – Jan 22, 2003 8 messages not in the archive ACPI

lore
Share & tools
lore

Robert Woerle Paceblade/Support

author
#0 lore raw

Hi

When i use mem=XXXM ACPI works fine , BUT i get cruel crashes when using
Sound and Ethernet or/and usb togethere ( they all have IRQ 5 ) one
device alone is fine .
When i dont use mem=XXXM i get an error at /drivers/acpi/tables/tb_get.c
BUT all the devices work fine , even together !!!
i tried also mem=XXXM + acpi=nopci which gives the same result : ACPI works and the devices crash together .... the machine is a Transmeta driving machine ...

So when i look at tb_get.c at the specific functions...
i see that it maps the pointers there according to a switch(adress->pointer) call ! Is possible that it normally uses the
case ACPI_LOGMODE_PHYSPTR:
stuff but when i use mem=XXX
it might use the other case statement:
case ACPI_LOGMODE_LOGPTR:

can i make a cruel hack around that that it always uses the working one ???

The aim is of course to get both variants together : ACPI working + DEvices not crashing !! Please advice !!!

---tb_get.c---- where it bails out--------

		status = acpi_os_map_memory (address->pointer.physical, sizeof 
(acpi_table_header),
				  (void **) &header);
		if (ACPI_FAILURE (status)) {
			ACPI_REPORT_ERROR (("Could not map memory at %8.8X%8.8X for length %X\n",
				ACPI_HIDWORD (address->pointer.physical),
				ACPI_LODWORD (address->pointer.physical),
				sizeof (acpi_table_header)));
			return_ACPI_STATUS (status);
		}

-

_____________________________________
*Robert Woerle
**Technical Support | Linux
PaceBlade Technology Europe SA*
phone: +49 89 552 99935
fax: +49 89 552 99910
mobile: +49 179 474 45 27
email: [email protected] <mailto:[email protected]>
web: http://www.paceblade.com
_____________________________________

-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

A message that is not in the archive (sent off-list or not yet received): [email protected]

#2 lore raw
Quoted 20 lines · Robert Woerle Paceblade/Support

Hi

When i use mem=XXXM ACPI works fine , BUT i get cruel crashes when using
Sound and Ethernet or/and usb togethere ( they all have IRQ 5 ) one
device alone is fine .
When i dont use mem=XXXM i get an error at /drivers/acpi/tables/tb_get.c BUT all the devices work fine , even together !!!
i tried also mem=XXXM + acpi=nopci which gives the same result : ACPI works and the devices crash together ....
the machine is a Transmeta driving machine ...

So when i look at tb_get.c at the specific functions... i see that it maps the pointers there according to a switch(adress->pointer) call !
Is possible that it normally uses the
case ACPI_LOGMODE_PHYSPTR:
stuff but when i use mem=XXX
it might use the other case statement:
case ACPI_LOGMODE_LOGPTR:

Quoted 2 lines

can i make a cruel hack around that that it always uses the working one ???

Can I see the dmesg without mem options?

Signature · 10 lines
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com

A message that is not in the archive (sent off-list or not yet received): 20030121131925.GB14593-j6u/t2rXLliUoIHC/UFpr9i2O/[email protected]

#4 lore raw
Quoted 19 lines · Ducrot Bruno
Quoted 16 lines · Robert Woerle Paceblade/Support

Hi

When i use mem=XXXM ACPI works fine , BUT i get cruel crashes when using Sound and Ethernet or/and usb togethere ( they all have IRQ 5 ) one device alone is fine . When i dont use mem=XXXM i get an error at /drivers/acpi/tables/tb_get.c
BUT all the devices work fine , even together !!!
i tried also mem=XXXM + acpi=nopci which gives the same result : ACPI works and the devices crash together ....
the machine is a Transmeta driving machine ...

So when i look at tb_get.c at the specific functions...
i see that it maps the pointers there according to a switch(adress->pointer) call ! Is possible that it normally uses the
case ACPI_LOGMODE_PHYSPTR:
stuff but when i use mem=XXX
it might use the other case statement:
case ACPI_LOGMODE_LOGPTR:

Quoted 8 lines
Quoted 4 lines

can i make a cruel hack around that that it always uses the working one ???

Can I see the dmesg without mem options?

what is also interessting is iomem

That is boot.msg without mem=XXX

Inspecting /boot/System.map
Loaded 17708 symbols from /boot/System.map.
Symbols match kernel version 2.4.19.
Loaded 47 symbols from 3 modules.
klogd 1.4.1, log source = ksyslog started.
<4>Linux version 2.4.19 (root@linux) (gcc version 3.2) #5 Sun Jan 12 09:46:26 CET 2003 <6>BIOS-provided physical RAM map:
<4> BIOS-e820: 0000000000000000 - 000000000009a800 (usable)
<4> BIOS-e820: 000000000009a800 - 0000000000100000 (reserved)
<4> BIOS-e820: 0000000000100000 - 000000000eefd000 (usable)
<4> BIOS-e820: 000000000eefd000 - 000000000ef00000 (ACPI NVS)
<4> BIOS-e820: 000000000ef00000 - 000000000f000000 (usable)
<4> BIOS-e820: 00000000fff80000 - 0000000100000000 (reserved)
<5>0MB HIGHMEM available.
<5>240MB LOWMEM available.
<7>ACPI: have wakeup address 0xc0001000
<6>Advanced speculative caching feature not present
<4>On node 0 totalpages: 61440
<4>zone(0): 4096 pages.
<4>zone(1): 57344 pages.
<4>zone(2): 0 pages.
<6>ACPI: RSDP (v000 PTLTD ) @ 0x000f81c0
<6>ACPI: RSDT (v001 PTLTD RSDT 01540.00005) @ 0x0eef9195
<6>ACPI: FADT (v001 TMETA PDB.ALI 01540.00005) @ 0x0eefcb64
<6>ACPI: BOOT (v001 PTLTD $SBFTBL$ 01540.00005) @ 0x0eefcbd8
<6>ACPI: DSDT (v001 TM PDBALI35 01540.00005) @ 0x00000000
<5>ACPI: BIOS passes blacklist
<4>Kernel command line: root=/dev/hda3 vga=791 init 3
<6>Initializing CPU#0
<4>Detected 599.225 MHz processor.
<4>Console: colour dummy device 80x25
<4>Calibrating delay loop... 1192.75 BogoMIPS
<6>Memory: 239880k/245760k available (1502k kernel code, 5460k reserved, 451k data, 116k init, 0k highmem) <6>Dentry cache hash table entries: 32768 (order: 6, 262144 bytes)
<6>Inode cache hash table entries: 16384 (order: 5, 131072 bytes)
<4>Mount-cache hash table entries: 4096 (order: 3, 32768 bytes)
<4>Buffer-cache hash table entries: 16384 (order: 4, 65536 bytes)
<4>Page-cache hash table entries: 65536 (order: 6, 262144 bytes)
<7>CPU: Before vendor init, caps: 0084893f 0081813f 0000000e, vendor = 7
<6>CPU: L1 I Cache: 64K (64 bytes/line), D cache 64K (32 bytes/line)
<6>CPU: L2 Cache: 512K (128 bytes/line)
<6>CPU: Processor revision 1.3.1.3, 600 MHz
<6>CPU: Code Morphing Software revision 4.2.7-8-278
<6>CPU: 20011004 02:04 official release 4.2.7#7
<7>CPU: After vendor init, caps: 0084893f 0081813f 0000000e 00000000
<5>CPU serial number disabled.
<7>CPU: After generic, caps: 0080893f 0081813f 0000000e 00000000
<7>CPU: Common caps: 0080893f 0081813f 0000000e 00000000
<4>CPU: Transmeta(tm) Crusoe(tm) Processor TM5600 stepping 03
<6>Checking 'hlt' instruction... OK.
<4>POSIX conformance testing by UNIFIX
<4>mtrr: v1.40 (20010327) Richard Gooch ([email protected]) <4>mtrr: detected mtrr type: none
<6>ACPI: Subsystem revision 20021022
<6>PCI: PCI BIOS revision 2.10 entry at 0xfd89e, last bus=0
<6>PCI: Using configuration type 1

<4>    ACPI-0135: *** Error: Could not map memory at 000000000EEF9195 for length 24
<4>    ACPI-0086: *** Error: Acpi_load_tables: Could not load RSDT: AE_NO_MEMORY
<4>    ACPI-0116: *** Error: Acpi_load_tables: Could not load tables: AE_NO_MEMORY

<3>ACPI: Unable to load the System Description Tables <6>PCI: Probing PCI hardware
<4>PCI: ACPI tables contain no PCI IRQ routing entries <4>PCI: Probing PCI hardware (bus 00)
..........................................................................
..........................................................................

AND here is boot.msg from the one with mem=238M

..........................................................................
Inspecting /boot/System.map
Loaded 17708 symbols from /boot/System.map.
Symbols match kernel version 2.4.19.
Loaded 47 symbols from 3 modules.
klogd 1.4.1, log source = ksyslog started.
<4>Linux version 2.4.19 (root@linux) (gcc version 3.2) #5 Sun Jan 12 09:46:26 CET 2003 <6>BIOS-provided physical RAM map:
<4> BIOS-e820: 0000000000000000 - 000000000009a000 (usable)
<4> BIOS-e820: 000000000009a000 - 0000000000100000 (reserved)
<4> BIOS-e820: 0000000000100000 - 000000000eefd000 (usable)
<4> BIOS-e820: 000000000eefd000 - 000000000ef00000 (ACPI NVS)
<4> BIOS-e820: 000000000ef00000 - 000000000f000000 (usable)
<4> BIOS-e820: 00000000fff80000 - 0000000100000000 (reserved)
<6>user-defined physical RAM map:
<4> user: 0000000000000000 - 000000000009a000 (usable)
<4> user: 000000000009a000 - 0000000000100000 (reserved)
<4> user: 0000000000100000 - 000000000ee00000 (usable)
<5>0MB HIGHMEM available.
<5>238MB LOWMEM available.
<7>ACPI: have wakeup address 0xc0001000
<6>Advanced speculative caching feature not present
<4>On node 0 totalpages: 60928
<4>zone(0): 4096 pages.
<4>zone(1): 56832 pages.
<4>zone(2): 0 pages.
<6>ACPI: RSDP (v000 PTLTD ) @ 0x000f81c0
<6>ACPI: RSDT (v001 PTLTD RSDT 01540.00005) @ 0x0eef9195
<6>ACPI: FADT (v001 TMETA PDB.ALI 01540.00005) @ 0x0eefcb64
<6>ACPI: BOOT (v001 PTLTD $SBFTBL$ 01540.00005) @ 0x0eefcbd8
<6>ACPI: DSDT (v001 TM PDBALI35 01540.00005) @ 0x00000000
<5>ACPI: BIOS passes blacklist
<4>Kernel command line: root=/dev/hda3 mem=238M vga=791
<6>Initializing CPU#0
<4>Detected 599.208 MHz processor.
<4>Console: colour dummy device 80x25
<4>Calibrating delay loop... 1192.75 BogoMIPS
<6>Memory: 237872k/243712k available (1502k kernel code, 5432k reserved, 451k data, 116k init, 0k highmem) <6>Dentry cache hash table entries: 32768 (order: 6, 262144 bytes)
<6>Inode cache hash table entries: 16384 (order: 5, 131072 bytes)
<4>Mount-cache hash table entries: 4096 (order: 3, 32768 bytes)
<4>Buffer-cache hash table entries: 16384 (order: 4, 65536 bytes)
<4>Page-cache hash table entries: 65536 (order: 6, 262144 bytes)
<7>CPU: Before vendor init, caps: 0084893f 0081813f 0000000e, vendor = 7
<6>CPU: L1 I Cache: 64K (64 bytes/line), D cache 64K (32 bytes/line)
<6>CPU: L2 Cache: 512K (128 bytes/line)
<6>CPU: Processor revision 1.3.1.3, 600 MHz
<6>CPU: Code Morphing Software revision 4.2.7-8-278
<6>CPU: 20011004 02:04 official release 4.2.7#7
<7>CPU: After vendor init, caps: 0084893f 0081813f 0000000e 00000000
<5>CPU serial number disabled.
<7>CPU: After generic, caps: 0080893f 0081813f 0000000e 00000000
<7>CPU: Common caps: 0080893f 0081813f 0000000e 00000000
<4>CPU: Transmeta(tm) Crusoe(tm) Processor TM5600 stepping 03
<6>Checking 'hlt' instruction... OK.
<4>POSIX conformance testing by UNIFIX
<4>mtrr: v1.40 (20010327) Richard Gooch ([email protected]) <4>mtrr: detected mtrr type: none
<6>ACPI: Subsystem revision 20021022
<6>PCI: PCI BIOS revision 2.10 entry at 0xfd89e, last bus=0
<6>PCI: Using configuration type 1
<6>ACPI: Interpreter enabled
<6>ACPI: Using PIC for interrupt routing
<6>ACPI: System [ACPI] (supports S0 S3 S4 S5)
<6>ACPI: PCI Root Bridge [PCI0] (00:00)
<4>PCI: Probing PCI hardware (bus 00)
<7>ACPI: PCI Interrupt Routing Table [\_SB_.PCI0._PRT]
<4>ACPI: PCI Interrupt Link [LNK1] (IRQs 5, disabled)
<4>ACPI: PCI Interrupt Link [LNK2] (IRQs *10)
<4>ACPI: PCI Interrupt Link [LNK3] (IRQs 10, disabled)
<4>ACPI: PCI Interrupt Link [LNK4] (IRQs 6 7 10 11, enabled at IRQ 5)
<4>ACPI: PCI Interrupt Link [LNK5] (IRQs 6 7 10 11, disabled)
<4>ACPI: PCI Interrupt Link [LNK6] (IRQs *10)
<4>ACPI: PCI Interrupt Link [LNK7] (IRQs *5)
<4>ACPI: PCI Interrupt Link [LNK8] (IRQs *5)
<4>ACPI: PCI Interrupt Link [LNKU] (IRQs *5)
<6>PCI: Probing PCI hardware
<4>ACPI: PCI Interrupt Link [LNK1] enabled at IRQ 5
<4>ACPI: PCI Interrupt Link [LNK3] enabled at IRQ 10
<4>ACPI: PCI Interrupt Link [LNK5] enabled at IRQ 11
<4>PCI: No IRQ known for interrupt pin A of device 00:0f.0 - using IRQ 255
<6>PCI: Using ACPI for IRQ routing

iomem without mem=XXXM

00000000-0009a7ff : System RAM
000a0000-000bffff : Video RAM area
000c0000-000c7fff : Video ROM
000d0000-000d0fff : Acer Laboratories Inc. [ALi] USB 1.1 Controller
000d0000-000d0fff : usb-ohci
000f0000-000fffff : System ROM
00100000-0eefcfff : System RAM
00100000-00277a45 : Kernel code
00277a46-002e887f : Kernel data
0eefd000-0eefffff : ACPI Non-volatile Storage < -----what is this ?????? there the mapping fails in tb_get.c ! is it mapped by himself already ??
0ef00000-0effffff : System RAM
10000000-10000fff : Texas Instruments PCI1410 PC card Cardbus Controller
10000000-10000fff : i82365
f8000000-fbffffff : Silicon Motion, Inc. SM720 Lynx3DM
f8200000-f89fffff : vesafb
fc000000-fc0fffff : Transmeta Corporation LongRun Northbridge fc100000-fc100fff : Lucent Microelectronics FW323
fc101000-fc101fff : Acer Laboratories Inc. [ALi] M5451 PCI AC-Link Controller Audio Device fc103000-fc1030ff : Realtek Semiconductor Co., Ltd. RTL-8139/8139C
fc103000-fc1030ff : 8139too
fff80000-ffffffff : reserved

and now iomem with mem=238M ( with other values ACPI fails the same like without the mem=XXX ! )

00000000-00099fff : System RAM <------ this has changed
000a0000-000bffff : Video RAM area
000c0000-000c7fff : Video ROM
000d0000-000d0fff : Acer Laboratories Inc. [ALi] USB 1.1 Controller
000d0000-000d0fff : usb-ohci
000f0000-000fffff : System ROM
00100000-0edfffff : System RAM <------------- this has changed
00100000-00277a45 : Kernel code
00277a46-002e887f : Kernel data
10000000-10000fff : Texas Instruments PCI1410 PC card Cardbus Controller
10000000-10000fff : i82365
f8000000-fbffffff : Silicon Motion, Inc. SM720 Lynx3DM
f8200000-f89fffff : vesafb
fc000000-fc0fffff : Transmeta Corporation LongRun Northbridge fc100000-fc100fff : Lucent Microelectronics FW323
fc101000-fc101fff : Acer Laboratories Inc. [ALi] M5451 PCI AC-Link Controller Audio Device fc103000-fc1030ff : Realtek Semiconductor Co., Ltd. RTL-8139/8139C
fc103000-fc1030ff : 8139too

Looks like ACPI prevents himself for copying into the "ACPI Non_volatile Storage " ???

Cheers Rob

-------------------------------------------------------
This SF.net email is sponsored by: Scholarships for Techies! Can't afford IT training? All 2003 ictp students receive scholarships. Get hands-on training in Microsoft, Cisco, Sun, Linux/UNIX, and more. www.ictp.com/training/sourceforge.asp

A message that is not in the archive (sent off-list or not yet received): [email protected]

#6 lore raw
Quoted 10 lines · Robert
Quoted 6 lines · Ducrot Bruno
Quoted 2 lines · Robert Woerle Paceblade/Support

Hi

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines

When i use mem=XXXM ACPI works fine , BUT i get cruel crashes when using Sound and Ethernet or/and usb togethere ( they all have IRQ 5 ) one device alone is fine .
When i dont use mem=XXXM i get an error at /drivers/acpi/tables/tb_get.c BUT all the devices work fine , even together !!!
i tried also mem=XXXM + acpi=nopci which gives the same result : ACPI works and the devices crash together ....
the machine is a Transmeta driving machine ...

Ook. Thank. Could you please send me in private your dsdt (apologize then if I got in the past..., but anyway that doesn't hurt :)
via something similar:

cat /proc/acpi/dsdt | gzip -c - | uuencode dsdt.gz | mail [email protected] -s '[ACPI] dsdt'

Thank in advance.

Signature · 11 lines
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.net email is sponsored by: Scholarships for Techies!
Can't afford IT training? All 2003 ictp students receive scholarships.
Get hands-on training in Microsoft, Cisco, Sun, Linux/UNIX, and more.
www.ictp.com/training/sourceforge.asp
#7 lore raw
In reply to a message not in the archive
Quoted 28 lines · Robert
Quoted 24 lines · Ducrot Bruno
Quoted 20 lines · Robert Woerle Paceblade/Support

Hi

When i use mem=XXXM ACPI works fine , BUT i get cruel crashes when using Sound and Ethernet or/and usb togethere ( they all have IRQ 5 ) one device alone is fine .
When i dont use mem=XXXM i get an error at /drivers/acpi/tables/tb_get.c BUT all the devices work fine , even together !!!
i tried also mem=XXXM + acpi=nopci which gives the same result : ACPI works and the devices crash together ....
the machine is a Transmeta driving machine ...

So when i look at tb_get.c at the specific functions... i see that it maps the pointers there according to a switch(adress->pointer) call !
Is possible that it normally uses the
case ACPI_LOGMODE_PHYSPTR:
stuff but when i use mem=XXX
it might use the other case statement:
case ACPI_LOGMODE_LOGPTR:

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

can i make a cruel hack around that that it always uses the working one ???

The problem is that the BIOS call to retrieve memory infos give 'false' information on how the memory controller is configured.
This is the messages you see as <4> BIOS-e820: ... (...)

and the code is in asm at very early stage of initialisation in arch/i386/boot/setup.S via int15 with param e820h.
In fact, the acpi tables should go to a 'special' type (the ACPI data one) but I don't see anything at all in the BIOS-e820 calls.

I want to verify now if your DSDT table have correct infos on how the memory controller have to be configured.

Signature · 11 lines
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.net email is sponsored by: Scholarships for Techies!
Can't afford IT training? All 2003 ictp students receive scholarships.
Get hands-on training in Microsoft, Cisco, Sun, Linux/UNIX, and more.
www.ictp.com/training/sourceforge.asp

A message that is not in the archive (sent off-list or not yet received): 20030121204235.GJ15023-j6u/t2rXLliUoIHC/UFpr9i2O/[email protected]

#9 lore raw
Quoted 33 lines · Ducrot Bruno
Quoted 25 lines · Robert
Quoted 21 lines · Ducrot Bruno
Quoted 16 lines · Robert Woerle Paceblade/Support

Hi

When i use mem=XXXM ACPI works fine , BUT i get cruel crashes when using Sound and Ethernet or/and usb togethere ( they all have IRQ 5 ) one device alone is fine . When i dont use mem=XXXM i get an error at /drivers/acpi/tables/tb_get.c
BUT all the devices work fine , even together !!!
i tried also mem=XXXM + acpi=nopci which gives the same result : ACPI works and the devices crash together ....
the machine is a Transmeta driving machine ...

So when i look at tb_get.c at the specific functions...
i see that it maps the pointers there according to a switch(adress->pointer) call ! Is possible that it normally uses the
case ACPI_LOGMODE_PHYSPTR:
stuff but when i use mem=XXX
it might use the other case statement:
case ACPI_LOGMODE_LOGPTR:

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

can i make a cruel hack around that that it always uses the working one ???

Quoted 9 lines

The problem is that the BIOS call to retrieve memory infos give 'false' information on how the memory controller is configured.
This is the messages you see as <4> BIOS-e820: ... (...)

and the code is in asm at very early stage of initialisation in arch/i386/boot/setup.S via int15 with param e820h.
In fact, the acpi tables should go to a 'special' type (the ACPI data one) but I don't see anything at all in the BIOS-e820 calls.

My first thought was that i do a cruel hack where in this stage (setup.S ) i do a minus - XXX at the total RAM I think the whole problem is that this is a Trasnmeta Crusoe machine at there are 16 MB RAM always taken by the CodeMorphing Technology and the BIOS doesnt remind that and doesnt report the correct values !!! I heard from other emails in this list that i might be possible to patch the DSDT to the correct values ??

That is what i thought from the beginning ... and since our BIOS Resources are very low these days , i would gladly get an Info how i could cruely workaround that for now to get everything done , since we plan our big kickout for the "Linux Tablet PC" for CEBIT(March ) this year together with Suse .
If we togther get it to run , you will get a small donation for sure !!!

Quoted 7 lines

I want to verify now if your DSDT table have correct infos on how the memory controller have to be configured.

-------------------------------------------------------
This SF.net email is sponsored by: Scholarships for Techies! Can't afford IT training? All 2003 ictp students receive scholarships. Get hands-on training in Microsoft, Cisco, Sun, Linux/UNIX, and more. www.ictp.com/training/sourceforge.asp

A message that is not in the archive (sent off-list or not yet received): [email protected]

#11 lore raw
Quoted 47 lines · Robert
Quoted 43 lines · Ducrot Bruno
Quoted 30 lines · Robert
Quoted 26 lines · Ducrot Bruno
Quoted 20 lines · Robert Woerle Paceblade/Support

Hi

When i use mem=XXXM ACPI works fine , BUT i get cruel crashes when using Sound and Ethernet or/and usb togethere ( they all have IRQ 5 ) one device alone is fine .
When i dont use mem=XXXM i get an error at /drivers/acpi/tables/tb_get.c BUT all the devices work fine , even together !!!
i tried also mem=XXXM + acpi=nopci which gives the same result : ACPI works and the devices crash together ....
the machine is a Transmeta driving machine ...

So when i look at tb_get.c at the specific functions... i see that it maps the pointers there according to a switch(adress->pointer) call !
Is possible that it normally uses the
case ACPI_LOGMODE_PHYSPTR:
stuff but when i use mem=XXX
it might use the other case statement:
case ACPI_LOGMODE_LOGPTR:

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

can i make a cruel hack around that that it always uses the working one ???

The problem is that the BIOS call to retrieve memory infos give 'false' information on how the memory controller is configured.
This is the messages you see as <4> BIOS-e820: ... (...)

Quoted 7 lines
Quoted 5 lines

and the code is in asm at very early stage of initialisation in arch/i386/boot/setup.S via int15 with param e820h.
In fact, the acpi tables should go to a 'special' type (the ACPI data one) but I don't see anything at all in the BIOS-e820 calls.

My first thought was that i do a cruel hack where in this stage (setup.S ) i do a minus - XXX at the total RAM

I think your first though is the correct to go. The problem why it doesn't work is that you have to recreate yourself the correct mapping, not only to reserve the top of mem as you have done.

For me, it sound like to be a critical bug in the firmware, that can have the potential to bug some propritary OS (not only free OS).

Signature · 11 lines
Ducrot Bruno

--  Which is worse:  ignorance or apathy?
--  Don't know.  Don't care.


-------------------------------------------------------
This SF.net email is sponsored by: Scholarships for Techies!
Can't afford IT training? All 2003 ictp students receive scholarships.
Get hands-on training in Microsoft, Cisco, Sun, Linux/UNIX, and more.
www.ictp.com/training/sourceforge.asp

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): 20030122142525.GA18109-j6u/t2rXLliUoIHC/UFpr9i2O/[email protected]

#15 lore raw
Quoted 19 lines

Well, normally bios inform the OS that the ACPI data region is for the usage of ACPI, and is located in the 'normal' RAM, no? Well, it's right also to thing that OSPM can tell then that
regions marked as 'ACPI data' can be used for normal RAM usage.
Perhaps Redmonds OS do that?

Well, well. I will look if we can (in general, not your perticuliar problem) use the ACPI data memory when we know that we can use them.

Back to your trouble:
Fortunately, we can add a e820 entry in arch/i386/kernel/setup.c Those, this prevent to make an cruel hack to arch/i386/boot/setup.S wich should be a little bit more harder..

Basically, I add a memory region marked for acpi data use, in sanitize_e820_map().

Quoted 9 lines

It's way more easy.. Could you please try without mem=xxx at all?

I see also that you use 2.4.19. Perhaps this patch will be OK for 2.4.19 but I am not sure at 100%.

Cheers,

Ok it boots up fine without mem=XX option i get ACPI to load ..... so gimme some time and i ll double check if the system is still crashing or not ...
but so far it looks nice ...

Rob

-------------------------------------------------------
This SF.net email is sponsored by: Scholarships for Techies! Can't afford IT training? All 2003 ictp students receive scholarships. Get hands-on training in Microsoft, Cisco, Sun, Linux/UNIX, and more. www.ictp.com/training/sourceforge.asp