copy_mount_options

copy_mount_options

[Rick Spanbauer]

Howdy - I am having a bit of a problem with a kernel port I've been working
on recently.  Linux is up and running to the initial sh prompt, networking/ethernet
works, and so on.  When I try a mount(), I am getting a kernel page fault.  Between
a few hours worth of debugging and some Googling around, I believe I understand
what is happening, ie when copy_mount_option is copying in from user space, it
seems to be running off the end of the data segment (code was apparently written with
the assumption that this is legal).  I can think of several different solutions, but
the question is this - is there some common, accepted practice solution to
this particular problem?  From searching the discussion groups, this class of problem
seems to be a known issue, but the arguments pro/con about what to do about
it never seem to converge :)  So before tracking off into the wilderness
on my own, I thought I might ask!  Tnx Rick Spanbauer

Re: copy_mount_options

[Jun Sun]

Rick Spanbauer wrote:

> Howdy - I am having a bit of a problem with a kernel port I've been working
> on recently.  Linux is up and running to the initial sh prompt, networking/ethernet
> works, and so on.  When I try a mount(), I am getting a kernel page fault.  Between
> a few hours worth of debugging and some Googling around, I believe I understand
> what is happening, ie when copy_mount_option is copying in from user space, it
> seems to be running off the end of the data segment (code was apparently written with
> the assumption that this is legal).  I can think of several different solutions, but
> the question is this - is there some common, accepted practice solution to
> this particular problem?  From searching the discussion groups, this class of problem
> seems to be a known issue, but the arguments pro/con about what to do about
> it never seem to converge :)  So before tracking off into the wilderness
> on my own, I thought I might ask!  Tnx Rick Spanbauer
> 


Congradulations!  You must have checked out the kernel from OSS CVS tree 
sometime in the past a couple of days.  There is a small time window when Ralf 
checked in the "correct epc for delay slot" fix with a typo and later got 
fixed again.  And coyp_mount_option() kernel fault was the symptom of the typo.

Get the latest code and give it a try.

Jun