On 1/13/21 7:45 PM, Randy Dunlap wrote:
On 1/13/21 2:29 AM, Hinko Kocevar wrote:
[I noticed this while working on PCI devices; not sure which kernel list would be best for this, though]
I noticed that my system shows address range for iomem and ioports as all 0. Sometimes (after a power cycle) the addresses are proper, albeit I have not been able to see that in a while now, after performing numerous reboots in the last coupe of days.
FWIW, I think the list of devices (names, count) looks the same in both cases. The system seems to work in both cases; at least I have not seen any complaints in kernel logs, OTOH, not sure what the error message would be.
What may be the reason for not getting the proper addresses listed?
config STRICT_DEVMEM
bool "Filter access to /dev/mem"
and
config IO_STRICT_DEVMEM
bool "Filter I/O access to /dev/mem"
so what are your CONFIG_STRICT_DEVMEM and CONFIG_IO_STRICT_DEVMEM set to?
Yeah, they're set. Did not notice that since I'm using some CentOS stock
config.
What do you see in /proc/iomem and /proc/ioports if you are admin/root?
That should be non-zero values.
You are right. Using sudo shows the values.
Thanks!
This likely poses any issues for userspace tools that would look at those /proc files, OTOH, I wonder if would kernel suffer in any way as well?
Yes, it could affect userspace.
Kernel version is 5.10.0 (pci git tree).
[dev@bd-cpu18 ~]$ cat /proc/iomem
00000000-00000000 : Reserved
00000000-00000000 : System RAM
00000000-00000000 : Reserved
00000000-00000000 : PCI Bus 0000:00
00000000-00000000 : Video ROM
00000000-00000000 : Reserved
00000000-00000000 : System ROM
00000000-00000000 : System RAM
00000000-00000000 : Kernel code
00000000-00000000 : Kernel rodata
00000000-00000000 : Kernel data
00000000-00000000 : Kernel bss
