[BUG] Red Hat Enterprise Linux AS 3 GM + kernel 2.6.0-test11 halt

[BUG] Red Hat Enterprise Linux AS 3 GM + kernel 2.6.0-test11 halt

[Zhu, Yi]

[-- Attachment #1: Type: text/plain, Size: 2974 bytes --]


Hi,

I installed Red Hat Enterprise Linux AS 3 GM on tiger4. Everything is
OK. Then I download the prestine 2.6.0-test11 kernel + ia64 patch 
http://www.kernel.org/pub/linux/kernel/ports/ia64/v2.6/linux-2.6.0-test1
1-ia64-031126.diff.gz

Compiled successfully. But when the new kernel is booted, it halted at 
"ACPI: Subsystem revision 20031002". The kernel is completely halted
without react any keyboard interupt. See my hand copy screen log in the
attachment. You can find my kernel .config in the attachment.

The same kernel (binary) can boot successfully on previous redhat AS
images and all the UL/SuSE images. I found there is a difference between
the new 
redhat AS 3 GM and previous ones is there is an additional
/boot/efi/Intel Fireware/fpswa.efi file!

My FPSWA version is also 1.18 from the boot screen. 

I read the Release Note and found below words
"Red Hat Enterprise Linux 3 for the Itanium 2 Architecture contains the
       fpswa.efi binary, which is an EFI (Extensible Firmware Interface)
       driver invoked by the operating system's kernel whenever the
processor
       issues FPSWA (Floating-Point Software Assistance) requests for
       floating-point operations that are not fully implemented in
hardware.
       The handler provides the result of the operation in these cases.
       Presently, Itanium processors issue FPSWA requests mainly when
very
       small numbers (denormalized, or tiny numbers) are consumed or
produced
       by floating-point instructions. To avoid the performance penalty
       caused by calls to the handler when tiny numbers are produced,
the
       flush-to-zero mode can be turned on, but at the cost of a
slightly
       less accurate computation.

       FPSWA is licensed, patented and copyrighted protected material by
       Intel or its supplier under the laws of the United states and
other
       countries and international treaty provisions. FPSWA is subject
to the
       applicable end user license agreement which can be accessed at
       http://www.redhat.com/licenses/."


The walk around way is as follow:

1. after the machine is booted, enter efi shell
2. remove fs0:\efi\"Intel Firmware"
3. save to NVRAM (*very important*)
4. exit efi shell and load your 2.6 kernel

Then you'll get the 2.6 kernel booted.

Everytime the machine is rebooted, I must redo step 3 above. Or the same
error will happen. :(

The default rh el kernel(Taroon 2.4.21-4.EL) boot ok without any
walkaround.

I believe this is a bug, either by EFI or by the 2.6 kernel.

Thanks,
-----------------------------------------------------------------
Opinions expressed are those of the author and do not represent Intel
Corp.

- Zhu Yi (Chuyee)

GnuPG v1.0.6 (GNU/Linux) http://cn.geocities.com/chewie_chuyee/gpg.txt
or $ gpg --keyserver wwwkeys.pgp.net --recv-keys 71C34820 1024D/71C34820
C939 2B0B FBCE 1D51 109A  55E5 8650 DB90 71C3 4820 

[-- Attachment #2: tiger4.txt --]
[-- Type: text/plain, Size: 887 bytes --]

Memory: 955392k/965136k available (8351k code, 21008k reserved, 2739k data, 2881 init)
Leaving McKinley Errata 9 workaround enabled
Calibrating delay loop... 1342.40 BogoMIPS
Dentry cache hash table entries: 65536 (order: 5, 524288 bytes)
Inode-cache hash table entries: 65536 (order: 5, 524288 bytes)
Mount-cache hash table entries: 1024 (order: 0, 16384 bytes)
POSIX conformance testing by UNIFIX
Boot processor id 0x0/0xc618
task migration cache decay timeout: 10 msecs.
Starting migration thread for cpu 0
Bringing up 1
Processor 49688/1 is spinning up...
Processor 0x1/0xc218 is stuck.
Bringing up 2
Processor 50200/2 is spinning up...
Processor 0x2/0xc418 is stuck.
Bringing up 3
Processor 49176/3 is spinning up...
Processor 0x3/0xc018 is stuck.
CPUS done 32
Total of 1 processors activated (1342.40 BogoMIPS).
NET: Registered protocol family 16
ACPI: Subsystem revision 20031002