On Sat, 22 Nov 2025, 20:08 Andreas Schwab, <[email protected]> wrote:
> On Nov 21 2025, Chet Ramey wrote: > > > Yes, this defeats the bash check. I tend to agree with kre that this > is a Linux kernel bug, but here we are. > For these virtual files where it *always* fails, yes that's a bug, but in general no, because… This can happen with any file (even "normal" regular ones) if another > process is modifying it just between the stat and the read calls. > (Agreed.) The correct way to check whether a whole file has been read is to keep reading until you get a short read. This implies, for example, that the buffer should be (at least) one byte larger than the expected size, so that you will normally get a short read straight away. But if you read a full buffer, just loop, reading until you get to the end of the file (expanding the buffer as needed). > + #if 0 > > if (nr != file_size) > > nr = -1; /* XXX - didn't get the whole file */ > > + #endif > This block is based on fragile assumptions and should be deleted (and replaced with something that will actually read the whole file). A short read is never an error, as you can see from the fact that errno > wasn't set. Yep. -Martin PS: mmap with a fallback to malloc+read ? >
