On 11/30/2017 08:43 AM, Michael S. Tsirkin wrote:
On Thu, Nov 30, 2017 at 07:41:59AM -0600, Eric Blake wrote:
The point of writing a macro embedded in a 'do { ... } while (0)'
loop is so that the macro can be used as a drop-in statement with
the caller supplying the trailing ';'. Although our coding style
frowns on brace-less 'if':
if (cond)
statement;
else
something else;
the use of do/while (0) in a macro is absolutely essential for the
purpose of avoiding a syntax error on the 'else' - but it only works
if there is no trailing ';' in the macro (as the ';' in the code
calling the macro would then be a second statement and cause the
'else' to not pair to the 'if').
Shouldn't matter if everyone puts the statements in {}, right?
Correct - where we follow our style, spurious ';' don't make a
difference (other than they might trigger a warning in a very particular
compiler). But it also makes our code harder to copy-and-paste into
other projects with a different style.
Many of the places touched in this code are examples of the ugly
bit-rotting debug print statements; cleaning those up is left as
a bite-sized task for another day.
Found mechanically via: $ git grep -B1 'while (0);' | grep -A1 \\\\
Signed-off-by: Eric Blake <ebl...@redhat.com>
We can't really rely on code still building for this to do the right
thing, can we?
Some of the uses that were changed are in dead #ifdefs for debugging
purposes - the only way to still compile is to turn on the debugging,
but that may fail to compile for other reasons (if the format strings
have bit-rotted, for example).
The only sure way to know that this did not break anything is to audit
that for every macro where I eliminated the ';', all callers of that
macro call 'foo();'. I did not perform that audit (the patch was
mechanical) - but am willing to do so and reply back with results if you
need the extra confidence.
I did my best to look for uses and I think it's OK, so
Reviewed-by: Michael S. Tsirkin <m...@redhat.com>
but I'm not merging this.
What tree should it go through instead? Does it need to be split along
maintainer boundaries?
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3266
Virtualization: qemu.org | libvirt.org
