Eric Blake <ebl...@redhat.com> writes: > On 06/22/2015 01:26 PM, Markus Armbruster wrote: >> This is particularly useful when we abort in error_propagate(), >> because there the stack backtrace doesn't lead to where the error was >> created. Looks like this: >> >> Unexpected error at /work/armbru/qemu/blockdev.c:322: >> qemu-system-x86_64: -drive if=none,werror=foo: 'foo' invalid >> write error action >> Aborted (core dumped) >> [Exit 134 (SIGABRT)] >> >> Note: to get this example output, I monkey-patched drive_new() to pass >> &error_abort to blockdev_init(). >> >> To keep the error handling boiler plate from growing even more, all >> error_setFOO() become macros expanding into error_setFOO_internal() >> with additional __FILE__, __LINE__ arguments. Not exactly pretty, but >> it works. > > I agree with Laszlo that adding __func__ to the mix also helps.
Okay, I'll give it a try. >> The macro trickery breaks down when you take the address of an >> error_setFOO(). Fortunately, we do that in just one place: qemu-ga's >> Windows VSS provider and requester DLL wants to call >> error_setg_win32() through a function pointer "to avoid linking glib >> to the DLL". Use error_setg_win32_internal() there. The use of the >> function pointer is already wrapped in a macro, so the churn isn't >> bad. >> >> Code size increases by some 14KiB for me (0.3%). Tolerable. Could be >> less if we passed relative rather than absolute source file names to >> the compiler. > > I also like it. > >> +#define error_setg(errp, fmt, ...) \ >> + error_setg_internal((errp), __FILE__, __LINE__, (fmt), ## __VA_ARGS__) >> +void error_setg_internal(Error **errp, const char *src, int line, >> + const char *fmt, ...) GCC_FMT_ATTR(4, 5); >> > >> +#define error_setg_errno(errp, os_error, fmt, ...) \ >> + error_setg_errno_internal((errp), __FILE__, __LINE__, (os_error), \ >> + (fmt), ## __VA_ARGS__) > > Nit - why the difference in \ alignment? I'm dense today... difference between where and where? > Nit - as used here, 'errp', 'fmt', and 'os_error' can be used > unambiguously; you don't need '(errp)' given the context of a > parenthesized comma-separated list (even if someone DID want to unusual > by passing in '(a,b)' with a comma operator for their 'errp' argument, > they'd have to supply the () because of the semantics of making the > macro call). I put parenthesis around macro parameters in the expansion pretty much unthinkingly, because thought is expensive :) > Nit - '## __VA_ARGS__' is a gcc-ism and not portable C99; but I think > clang supports it, and we don't really care about other compilers at the > moment. At any rate, we already use it elsewhere in qemu.git. Correct. For what it's worth, ./HACKING recommends it.
