On 26 September 2016 at 16:36, Riku Voipio <riku.voi...@linaro.org> wrote: > On 27 September 2016 at 00:08, Peter Maydell <peter.mayd...@linaro.org> wrote: >> Do you have some examples of the false positives you want >> to suppress here? For new code I would hope that we can >> handle host-arch-specifics by having new files (or just >> new #defines etc) in linux-user/host/$ARCH/ rather than >> inline #ifdeffery in the main files. > > One example from your patch: > > https://lists.gnu.org/archive/html/qemu-devel/2016-09/msg05650.html > > And another from Laurent: > > https://lists.gnu.org/archive/html/qemu-devel/2016-09/msg06486.html > > Every new syscall will comes with "#ifdef TARGET_NR_foo and > defined(__NR_foo)", while host/target combos catch up. Now, most > TARGET_NR_foo's are needed only for unicore32, but the __NR_foo > defines will be needed for a very long time.
Oh, I see; I don't think of the __NR_foo as being "architecture specific". I think we'd be better off specifically whitelisting those in checkpatch rather than turning off the whole check for linux-user. thanks -- PMM
