2.6.15-13-amd64-k8 praise and issues

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

2.6.15-13-amd64-k8 praise and issues

Anthony Durity
Hello there,

    First off - great work. Thanks for the shiny new kernel in Dapper.

    If you could possibly grab the rt2570 USB nightly CVS tarball rather
than the Latest BETA rt2570 driver that would be great. The CVS has a
couple of nice fixes and has zero code churn so you can trust it. All
their work is being concentrated elsewhere (rt2x00) so you can snag this
without a worry. Specifically strength reports x/100 rather than 0.

    A couple of amd-64 config things. CONFIG_X86_MCE_INTEL is set to y,
should be n? CONFIG_LOG_BUF_SHIFT is set to 14, should be 16 - kernel
config help says 14 is i386, 16 is x86_64. dmesg is truncated at the
beginning otherwise and this is quite annoying.

    The new experimental bcm43xx won't even report an access point
FF:FF:FF:FF:FF:FF let alone connect and hangs my laptop on rmmod. My
chip is a BCM4318 [AirForce One 54g].

    My machine won't boot this kernel unless I boot with noapic. Of
course I lose my apic and I do so love my apic. I have had this issue so
many times with so many machines that I think maybe you should have this
as standard across the board, live CD, install CD, you name it. I think
other distros do this. Is there an absolute reason we cannot?

    Another way to get around the non-booting unhappiness is (on my
compaq r4218ea) to pass acpi=debug or acpi=strict. When I do this
however I get
[   80.275360] APIC error on CPU0: 00(40)
[  133.578525] APIC error on CPU0: 40(40)
in my dmesg output. I have done an amount of Googling and cannot figure
out what it is. Any suggestions?

    I have attached my hardware info. I have no idea what the unknown
devices in lspci are, probably nothing.

Thanks all and respect!

kernel-team mailing list
[hidden email]

lscpi.txt (2K) Download Attachment
interrupts-apic.txt (650 bytes) Download Attachment
lshw-1.txt (14K) Download Attachment