Re: 2x procfs bug with 2.4.23

Re: 2x procfs bug with 2.4.23

[Len Brown]

Auke,
Problem #2 should be fixed by the patch here:
http://bugzilla.kernel.org/show_bug.cgi?id=991
Please let me know if it works for you.

thanks,
-Len


On Sun, 2003-12-14 at 18:51, Auke Kok wrote:
> Hi,
> 
> I have encountered 2 different procfs bugs with 2.4.23, where procfs
> seems to be running a possible race, the other with ACPI subdirs being
> created in /proc instead of /proc/acpi, possibly leading to more
> problems.
> 
> both kernels run the official grsec patchset, so I forward them to
> 
> 1) procfs sys shows alternatingly up as "file", "socket", and
> "directory"
> 
> root@server /proc # ls -ld sys
> srwxrwxrwx   10 root     root            0 2003-12-15 00:38 sys
> root@server /proc # ls -ld sys
> -r--r--r--   10 root     root            0 2003-12-15 00:38 sys
> root@server /proc # ls -ld sys
> prw-------   10 root     root            0 2003-12-15 00:41 sys
> root@server /proc # ls -ld sys
> prw-------   10 root     root            0 2003-12-15 00:41 sys
> root@server /proc # ls -ld sys
> srwxrwxrwx   10 root     root            0 2003-12-15 00:41 sys
> root@server /proc # cd sys
> root@server /proc/sys # ls -ld .
> srwxrwxrwx   10 root     root            0 2003-12-15 00:41 .
> root@server /proc/sys # pwd
> /proc/sys
> root@server /proc/sys # ls
> abi  debug  dev  fs  kernel  net  proc  vm
> 
> etc... very very annoying since this box needs ip_forward enabled
> after
> boot to function!
> 
> The other box has an unsupported ACPI core:
> 
> [from dmesg]
> ACPI: Unable to locate RSDP
> ACPI: Subsystem revision 20031002
> PCI: PCI BIOS revision 2.10 entry at 0xf7ec0, last bus=0
> PCI: Using configuration type 1
> ACPI: System description tables not found
>      ACPI-0084: *** Error: acpi_load_tables: Could not get RSDP,
> AE_NOT_FOUND
>      ACPI-0134: *** Error: acpi_load_tables: Could not load tables:
> AE_NOT_FOUND
> ACPI: Unable to load the System Description Tables
> PCI: Probing PCI hardware
> PCI: ACPI tables contain no PCI IRQ routing entries
> 
> it's a genuine pII-mmx, so I don't care about the ACPI. However, the
> /proc/acpi directory isn't created but the subdirs for enabled
> components are, which could result in confusion/problems:
> 
> ~ # ls -dF /proc/[a-zA-Z_-]*
> /proc/ac_adapter/  /proc/interrupts  /proc/partitions
> /proc/battery/     /proc/iomem       /proc/pci
> /proc/bus/         /proc/ioports     /proc/processor/
> /proc/button/      /proc/irq/        /proc/self@
> /proc/cmdline      /proc/kmsg        /proc/slabinfo
> /proc/cpuinfo      /proc/ksyms       /proc/stat
> /proc/crypto       /proc/loadavg     /proc/swaps
> /proc/devices      /proc/locks       /proc/sys/
> /proc/dma          /proc/lvm/        /proc/sysvipc/
> /proc/driver/      /proc/mdstat      /proc/thermal_zone/
> /proc/execdomains  /proc/meminfo     /proc/tty/
> /proc/fan/         /proc/misc        /proc/uptime
> /proc/filesystems  /proc/modules     /proc/version
> /proc/fs/          /proc/mounts@
> /proc/ide/         /proc/net/
> 
> [note the thermal_zone... it should not be there I think ;^)]
> 
> 
> -- not subscribed, reply to sofar <at> lunar.linux <dot> org
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe
> linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 
> 

2x procfs bug with 2.4.23

[Auke Kok]

Hi,

I have encountered 2 different procfs bugs with 2.4.23, where procfs 
seems to be running a possible race, the other with ACPI subdirs being 
created in /proc instead of /proc/acpi, possibly leading to more problems.

both kernels run the official grsec patchset, so I forward them to

1) procfs sys shows alternatingly up as "file", "socket", and "directory"

root@server /proc # ls -ld sys
srwxrwxrwx   10 root     root            0 2003-12-15 00:38 sys
root@server /proc # ls -ld sys
-r--r--r--   10 root     root            0 2003-12-15 00:38 sys
root@server /proc # ls -ld sys
prw-------   10 root     root            0 2003-12-15 00:41 sys
root@server /proc # ls -ld sys
prw-------   10 root     root            0 2003-12-15 00:41 sys
root@server /proc # ls -ld sys
srwxrwxrwx   10 root     root            0 2003-12-15 00:41 sys
root@server /proc # cd sys
root@server /proc/sys # ls -ld .
srwxrwxrwx   10 root     root            0 2003-12-15 00:41 .
root@server /proc/sys # pwd
/proc/sys
root@server /proc/sys # ls
abi  debug  dev  fs  kernel  net  proc  vm

etc... very very annoying since this box needs ip_forward enabled after 
boot to function!

The other box has an unsupported ACPI core:

[from dmesg]
ACPI: Unable to locate RSDP
ACPI: Subsystem revision 20031002
PCI: PCI BIOS revision 2.10 entry at 0xf7ec0, last bus=0
PCI: Using configuration type 1
ACPI: System description tables not found
     ACPI-0084: *** Error: acpi_load_tables: Could not get RSDP, 
AE_NOT_FOUND
     ACPI-0134: *** Error: acpi_load_tables: Could not load tables: 
AE_NOT_FOUND
ACPI: Unable to load the System Description Tables
PCI: Probing PCI hardware
PCI: ACPI tables contain no PCI IRQ routing entries

it's a genuine pII-mmx, so I don't care about the ACPI. However, the 
/proc/acpi directory isn't created but the subdirs for enabled 
components are, which could result in confusion/problems:

~ # ls -dF /proc/[a-zA-Z_-]*
/proc/ac_adapter/  /proc/interrupts  /proc/partitions
/proc/battery/     /proc/iomem       /proc/pci
/proc/bus/         /proc/ioports     /proc/processor/
/proc/button/      /proc/irq/        /proc/self@
/proc/cmdline      /proc/kmsg        /proc/slabinfo
/proc/cpuinfo      /proc/ksyms       /proc/stat
/proc/crypto       /proc/loadavg     /proc/swaps
/proc/devices      /proc/locks       /proc/sys/
/proc/dma          /proc/lvm/        /proc/sysvipc/
/proc/driver/      /proc/mdstat      /proc/thermal_zone/
/proc/execdomains  /proc/meminfo     /proc/tty/
/proc/fan/         /proc/misc        /proc/uptime
/proc/filesystems  /proc/modules     /proc/version
/proc/fs/          /proc/mounts@
/proc/ide/         /proc/net/

[note the thermal_zone... it should not be there I think ;^)]


-- not subscribed, reply to sofar <at> lunar.linux <dot> org