Messages in this thread | | | From | Robin Rosenberg <> | Subject | Re: Lockless file reading | Date | Thu, 28 Aug 2003 12:54:11 +0200 |
| |
Aren't Linux files also streams. Writing to a stream in sequence should update the stream in sequence, or it wouldn't be a stream, would it?
-- robin
onsdagen den 27 augusti 2003 14.37 skrev Timo Sirainen: > (Maybe a bit off topic, but I couldn't get answers elsewhere and it is > about kernel behaviour) > > The question is what can happen if I read() a file that's being > simultaneously updated by a write() in another process? If I'm writing > 123 over XXX, is it possible that read() returns 1X3 in some conditions? > Is the behaviour filesystem specific? Any idea about other operating > systems? > > I'm thinking about implementing lockless file reads that work like: > > void write_data(int fd, off_t offset, void *data, size_t size) { > lock_file(fd); > pwrite(fd, data, size, offset); // or writev() or copy+single pwrite() > pwrite(fd, data, size, offset + size); > unlock_file(fd); > } > > void read_data(int fd, off_t offset, void *data, size_t size) { > unsigned char buf{size*2]; > for (;;) { > pread(fd, buf, size*2, offset); // or shared mmap()ed access > if (memcmp(buf, buf+size, size) == 0) break; > usleep(10); > } > memcpy(data, buf, size); > } > > The only case when I see that it would break is if we write "1212" but > pread() sees "1X1X" or "X2X2" there. > > > - > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > Please read the FAQ at http://www.tux.org/lkml/
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |