On 09/19/2017 04:06 AM, Paolo Bonzini wrote: > On 13/09/2017 21:08, John Snow wrote: >> >> >> On 09/13/2017 06:21 AM, Thomas Huth wrote: >>> Remove the unnecessary home-grown redefinition of the assert() macro here, >>> and remove the unusable debug code at the end of the checkpoint() function. >>> The code there uses assert() with side-effects (assignment to the "mapping" >>> variable), which should be avoided. Looking more closely, it seems as it is >>> apparently also only usable for one certain directory layout (with a file >>> named USB.H in it) and thus is of no use for the rest of the world. >>> >>> Signed-off-by: Thomas Huth <th...@redhat.com> >> >> Farewell, bitrot code. >> >> Reviewed-by: John Snow <js...@redhat.com> >> >> Out of curiosity, I wonder ... >> >> jhuston@probe (foobar) ~/s/qemu> git grep '#if 0' | wc -l >> 320 > > > $ git grep -c '#if 0' | sort -k2 --field-separator=: -n > ... > hw/net/eepro100.c:21 > target/ppc/cpu-models.h:76 > > whoa :) >
Wonder if '#if 0' should be against the style guide / in checkpatch. Conditional compilations should at least be contingent on some named variable/condition. (Probably super ideally all conditional compilations correlate directly to a configure variable...) --js