On Fri, 26 Apr 2024, Michael Schmitz wrote:
Similar tests with start address five or six bytes before the start of the unmapped page, and corresponding path length to be returned by getcwd have shown no more Oops on 030 using the attached corrected version of my patch. Please give that some testing if you can, and (hoping it won't show any new faults on 040) I'll post another version of the series with your Tested-by added.
I will test it tomorrow. I expect that a NOP is needed at the end of __clear_user. BTW, since you're changing this line, I think there should be a tab here rather than two spaces: + "11: "MOVES".b %3,(%2)+\n"
