On Tue, Mar 23, 2010 at 8:45 PM, John David Anglin <dave@xxxxxxxxxxxxxxxxxx> wrote: >> The execve_wrapper macro loads thread->task like this: >> LDREG TI_TASK-THREAD_SZ_ALGN-FRAME_SIZE(%r30),%r1 >> >> This assumes that at the point the macro is called that there is only >> THREAD_SZ_ALGN+FRAME_SIZE between %r30 and the starting value of %r30 >> (cr30). >> >> However, in the syscall entry path I see (syscall.S): >> ldo THREAD_SZ_ALGN+FRAME_SIZE(%r30),%r30 /* set up kernel stack */ >> ... >> STREGM %r1,FRAME_SIZE(%r30) /* save r1 (usp) here for now */ >> >> Which means that between %r30 and the start there is actually >> THREAD_SZ_ALGN+2*FRAME_SIZE. > > This is wrong as there is a compensating LDREGM: > > LDREGM -FRAME_SIZE(%r30), %r2 /* get users sp back */ Thanks, I missed the `M` and thought it was just loading from a negative offset. >> or heaven forbid the easier to read: >> >> mfctl %cr30,%r1 /* get task ptr in %r1 */ >> LDREG TI_TASK(%r1),%r1 > > The above looks safer... It is, but a little bit slower, does it matter? This change would make execve_wrapper independent of r30's value. I think for now I'll just document the requirement. Cheers, Carlos. -- To unsubscribe from this list: send the line "unsubscribe linux-parisc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
