On 31 July 2012 19:38, Christoffer Dall <c.dall@xxxxxxxxxxxxxxxxxxxxxx> wrote: > On Fri, Jul 13, 2012 at 9:57 AM, Peter Maydell <peter.maydell@xxxxxxxxxx> wrote: >> On 13 July 2012 05:43, Rusty Russell <rusty.russell@xxxxxxxxxx> wrote: >>> 2) Why do we trap c11 accesses in HSTR? AFAICT there's nothing there... >> >> Relic from A9 support, maybe? We should only be setting HSTR.15 >> now. >> > I set this bit since the ARM arm specifies accesses to it as > unpredictable, and I wasn't sure about the ramifications on the host > if the guest accesses c11 accesses from PL1 and wanted to make sure > there were none. Is this in deed the case for undefined behavior UNPREDICTABLE has a specific meaning for ARM (see the glossary in the ARM ARM). In particular "UNPREDICTABLE behavior must not perform any function that cannot be performed at the current or lower level of privilege using instructions that are not UNPREDICTABLE". So leaving c11 untrapped isn't letting the guest do anything it couldn't do anyway. (UNDEFINED is something else, meaning "generates an undefined instruction exception".) -- PMM _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
