On Sun, Oct 14, 2012 at 1:11 PM, Peter Maydell <peter.mayd...@linaro.org> wrote: > This patch series introduces a new logging category to > qemu_log_mask(), for reporting situations which the device > model can cope with but which indicate bugs in the guest. > The most common cause of these is "guest tried to access > a register which doesn't exist". At the moment device models > use an inconsistent mix of "silently ignore", "hw_error(), > killing qemu" and "direct print to stderr" for reporting > this kind of event. > > Patches 2-7 are a random selection of ARM devices which I've > switched to using the new LOG_GUEST_ERROR (or the existing > LOG_UNIMP where that is more appropriate). There are obviously > more conversions that could be made; this is just a sample > to demonstrate the utility of the log category.
The conversion is not entirely trivial, since hw_error() does not return but qemu_log_mask() does. Your changes look OK. > > One question that comes to mind -- should we include qemu-log.h > in qemu-common.h (or some other common header?) rather than > having to include qemu-log.h directly in lots of device model > source files? Yes. > > Peter Maydell (7): > qemu-log: Add new log category for guest bugs > hw/pl181: Use LOG_UNIMP and LOG_GUEST_ERROR > hw/pl041: Use LOG_UNIMP > hw/pl190: Use LOG_GUEST_ERROR > hw/pl011: Use LOG_UNIMP and LOG_GUEST_ERROR > hw/pl022: Use LOG_UNIMP and LOG_GUEST_ERROR > hw/pl031: Use LOG_GUEST_ERROR > > hw/pl011.c | 12 ++++++++---- > hw/pl022.c | 9 ++++++--- > hw/pl031.c | 17 +++++++++++------ > hw/pl041.c | 6 ++++-- > hw/pl181.c | 19 +++++++++++-------- > hw/pl190.c | 7 +++++-- > qemu-log.c | 3 +++ > qemu-log.h | 1 + > 8 files changed, 49 insertions(+), 25 deletions(-) > > -- > 1.7.11.4 >