crash
crash copied to clipboard
crash-utility
missing (un)xlate_dev_mem_ptr exports on s390x
Loading crash module on s390x these messages are seen:
[ 129.794298] crash: loading out-of-tree module taints kernel. [ 129.795196] crash: Unknown symbol unxlate_dev_mem_ptr (err 0) [ 129.795379] crash: Unknown symbol xlate_dev_mem_ptr (err 0)
On most architectures xlate_dev_mem_ptr is #defined to __va but on s390x it is implemented as actual function.
----- Original Message -----
Loading crash module on s390x these messages are seen:
[ 129.794298] crash: loading out-of-tree module taints kernel. [ 129.795196] crash: Unknown symbol unxlate_dev_mem_ptr (err 0) [ 129.795379] crash: Unknown symbol xlate_dev_mem_ptr (err 0)
On most architectures xlate_dev_mem_ptr is #defined to __va but on s390x it is implemented as actual function.
I never really tested the sample driver on s390x, but I see that both xlate_dev_mem_ptr() and unxlate_dev_mem_ptr() are EXPORT_SYMBOL_GPL() in RHEL, and it appears that's not case with other (upstream) kernels. So the driver is apparently unusable on such kernels.
I don't recall the exact reason why IBM asked us to change the use of the architecture-neutral map_virtual() function, which uses kmap(). You could try excluding the CONFIG_S390 segment and see what happens.
Dave
-- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/crash-utility/crash/issues/19
I can export the functions with no problem.
The point here is that either the driver should be fixed to work with non-RH kernels or the export submitted upstream.
----- Original Message -----
I can export the functions with no problem.
The point here is that either the driver should be fixed to work with non-RH kernels or the export submitted upstream.
That driver is just a last-hope template for whomever wants to play with it, so if IBM wants to do so, they certainly can. I'm not particularly interested in doing so, and with the security folks constantly on the war-path against allowing physical memory access, it's doubtful it would even be accepted.
-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/crash-utility/crash/issues/19#issuecomment-322826772