On 07/02/14 03:54, Christophe Lyon wrote: > On 6 February 2014 10:49, Yury Gribov <y.gri...@samsung.com> wrote: >> Kugan wrote: >>>> Ok if no regressions. >>> >>> Tested it on qemu for arm-none-linux-gnueabi and there is no new >>> regressions. I am sorry I didn't mention it when I posted the patch. >> >> Commited in r207533. Thanks! >> > > Hi, > > As can be seen here: > http://cbuild.validation.linaro.org/build/cross-validation/gcc/207533/report-build-info.html > this has caused some regressions on armv5t targets.
Thanks for spotting this. armv5t disables unaligned access by default. I am looking into it. > > Additionally, are you sure testing this kind of thing on QEMU > sufficient? Does it really handle unaligned accesses as bus errors? > I did test this patch on a Chromebook where I reproduced the bus error with an older version of trunk and verified it was fixed. I then tested the patch for armv7-a with the latest trunk on qemu. I am not sure if qemu models the bus errors. Thanks, Kugan > Christophe. >