On Fri, 13 Nov 2020 13:03:05 -0500 Steven Rostedt <rost...@goodmis.org> wrote:
> On Fri, 13 Nov 2020 09:57:54 -0800 > Linus Torvalds <torva...@linux-foundation.org> wrote: > > > On Fri, Nov 13, 2020 at 9:54 AM Steven Rostedt <rost...@goodmis.org> wrote: > > > > > > The issue is that grub padded the end of the ramdisk after loading it > > > into memory. I'm not sure how the bootconfig tool can fix this. Perhaps > > > make sure the ram disk size is 4 bytes aligned? > > > > Exactly. Since - as far as I can tell - the _only_ thing that actually > > generates that BOOTCONFIG_MAGIC marker is the bootconfig tool, you > > control the vertical and the horizontal. No need for some "heuristic" > > and searching for things. > > > > And then that thing needs to be documented, so that if somebody else > > starts generating BOOTCONFIG_MAGIC markers, we have a hard rule in > > place that "look, the bootconfig is always aligned". > > > > Might as well align it more than 4 bytes while at it and make it even > > stricter. > > > > OK, yes I agree with this. OK, but note that the initrd_end can be modified by the bootloader anyway. The bootloader can pass any "size" bigger than actual initramfs size because initramfs is a cpio file which has a "TRAILER!!!" magic as the end of file. This means kernel can ignore or use (as the bootconfig does) for a tailing data storage. But I agree that we need to document it, so that anyone can refer the data format. > Masami, can you send a patch to fix the bootconfig tool to make sure > that when it appends to the initrd that it makes sure the file size is > aligned. Would 32 bytes be big enough for an alignment? OK, it is easy to me to update bootconfig tool to align up the total size to 32bytes, but I think 4 bytes align is OK if we document it. Without documentation, no one in the bootloader decides what is the correct format. Thank you, -- Masami Hiramatsu <mhira...@kernel.org>
