> The problem is that having buffers doesn't necessarily always mean that
> they are valid, nor that _all_ of them are valid.
>
> Also, if the ramdisk "readpage" code is wrong, then so is the
> "prepare_write" code. They share the same logic, which basically says
> that "if the page isn't up-to-date, then it is zero". Which is always
> true for normal read/write accesses, but as you found out it's not true
> when parts of the page have been accessed by filesystems through the
> buffers.
AFAICS, it's nastier than that. What's to stop buffer_heads to be
freed under memory pressure?
-
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/