On Wed, 2011-02-02 at 17:09 +0900, Ajeet Yadav wrote:
> If I see the current sigfault, its easy to fix adding one more patch
> to xfsprogs.
. . .
> But when I start reviewing the complete project w.r.t read() /
> read64() / write() / write64() more importantly libxfs_readbufr() /
> I find error handing is broken at may places and I get my self lost in
> m^n complexity also errno is lost.. therefore caller cannot examine
> the exact error,
> Back again I think, What if I exit on error ? Does xfsprogs uses
> read() / write() error as a part of its functionality, for example
> does xfs_repair uses these errors as a part of repair funtionality.
. . .
I think these are good observations and questions.
This doesn't answer your questions, but I want to mention to
you that libxfs is just about to have a major update that
affects lots of files. I expect the update will *not*
address the issues you point out here, but I guess what I'd
like to do is get us moved to the updated code before trying
to address problems that might be widespread like this.
I hope to complete this update process soon, but I expect
it will be a week or more before that happens.