RE: [PATCH 1/2] iomap: swap: print warning for unaligned swapfile

From: Sukrit.Bhatnagar@xxxxxxxx
Date: Mon May 27 2024 - 07:32:32 EST


Hi Chris,

On 2024-05-23 03:02, Chris Li wrote:
> Hi Sukrit,
>
> It seems that you need the swap file block start address to read the
> swap file headers.
> This warning still requires the user to read the dmesg. The kernel
> still does not have the swapfile header at resume. In other words, it
> does not fix the issue.

This was not intended to be a fix.

I had created this patch thinking that adding an actual fix for this might
be non-trivial and may not be desirable to everyone, especially when
swapfile+hibernation is not commonly used.

> I don't know the suspend/resume code enough, will adding recording the
> physical start address of the swapfile in swap_info_struct help you
> address this problem? The suspend code can write that value to
> "somewhere* for resume to pick it up.
>
> Let's find a proper way to fix this issue rather than just warning on it.

Adding a new member in swap_info_struct for the physical block offset
will help in the fix, I think. It can even be enclosed in #ifdef HIBERNATION
if nothing else needs that value.
This value can be checked by hibernate[1] as:
sis->start_offset == first_se(sis)->start_block

Another possible solution might be to add a new swap flag like SWP_SUSP
or SWP_HIBERNATE and set it only when we don't have this rounding up
issue.

Since we will boot the kernel normally first and later switch to our
pre-hibernate kernel after image load from swap, the value cannot be
set inside the kernel memory.
We pass the start block address in kernel commandline parameter
(resume_offset) for next boot.

[1]: The hibernate swapfile checking code is the function
swap_type_of() in mm/swapfile.c

--
Sukrit