On 02.11.2015 14:09, Peter Maydell wrote: > On 21 October 2015 at 19:15, Sergey Fedorov <[email protected]> wrote: >> On 16.10.2015 16:58, Peter Maydell wrote: >>> From: Sergey Fedorov <[email protected]> >>> >>> A QEMU breakpoint match is not definitely an architectural breakpoint >>> match. If an exception is generated unconditionally during translation, >>> it is hardly possible to ignore it in the debug exception handler. >>> >>> Generate a call to a helper to check CPU breakpoints and raise an >>> exception only if any breakpoint matches architecturally. >>> >>> Signed-off-by: Sergey Fedorov <[email protected]> >>> Reviewed-by: Peter Maydell <[email protected]> >>> Signed-off-by: Peter Maydell <[email protected]> >>> --- >> It turns out that this change introduced an issue which can be >> illustrated by the following test: >> I think we could fix this problem by cleaning up DISAS_UPDATE usage in >> target-arm/translate.c and implementing PC update as in >> target-arm/translate-a64.c. I could prepare a patch for that. >> >> Another problem, I think, is that we should somehow restore the CPU >> state before raising an exception from check_breakpoints() helper. But >> so far I have no idea how to fix this... > Hi, Sergey -- how are you doing with the fix for this? It would > be good to get it in and tested soon, because hardfreeze is next > week. > > I've also had a report that this patch broke gdbstub single-stepping, > which might be the same underlying cause.
Hi Peter, The patch for DISAS_UPDATE is almost ready. Basically, all I need is to prepare a commit message. But I'm not sure how to deal with CPU state restoring issue. Also it's a strange thing about gdbstub single-stepping I'm going to look at it. Best, Sergey
