[Bug 271766] dumpfs fails on geli devices: cylinder group checks failed

2023-06-05 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=271766

--- Comment #13 from Mark Johnston  ---
(In reply to Andrew "RhodiumToad" Gierth from comment #9)
I think there's a bug in the handling of the end-of-cursor condition in
crypto_cursor_segment().  Some cursor types handle it properly (e.g., mbufs),
others do not (e.g., vm_pages, used by GELI).

This might be unrelated to the crash, but I think it's a real bug.  I'll try
the repro shortly.

-- 
You are receiving this mail because:
You are the assignee for the bug.


[Bug 271766] dumpfs fails on geli devices: cylinder group checks failed

2023-06-05 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=271766

--- Comment #14 from Mark Johnston  ---
Created attachment 242627
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=242627&action=edit
address the panic

This patch fixes the crash that I can reproduce on stable/13 using the test
program.  It does not appear to fix the dumpfs failure.

On main, I cannot trigger a panic, but the test program reports discrepancies
(which it does not do with a patched stable/13).  I cannot reproduce the dumpfs
error on main.

I'm quite sure that the patch is required on main as well, i.e., the panic is
not specific to stable/13.

-- 
You are receiving this mail because:
You are the assignee for the bug.