Raymund Will changed bug 996088
What Removed Added
Flags   needinfo?(stuart@stella-maris.org.uk)

Comment # 6 on bug 996088 from
Please refrain from UEFI/NVRAM experiments on self-paid hardware, it may
well break... (does anybody remember Samsung laptops or intel motherboards?)

Looking at the comments above, brings me to the following "remarks":
1. your system apparently has trouble with garbage collection in the
   NVRAM management.  That "no space left on device" in fact refers
   to NVRAM, which "suffered" one addition of a variable (the refreshed
   boot entry #0000), followed by one replacement (BootOrder +1), plus one
   removal (the now obsolete old boot entry #0003) and then another
   replacement (BootOrder -1) during that grub2-update.
   Obviously the last step failed.
2. This state persists until after a reset/reboot. UEFI garbage-collection
   is performed and write operations can succeed again -- unless a buggy
   firmware and careless OS/apps conspire and leave less free NVRAM available
   than necessary, in which case all NVRAM write operations fail (even vital
   ones by UEFI itself).  Usually an external programmer and firmware+
   documentation unavailable to mere mortals) is needed to revive such
   a motherboard. (IIRC linux has patches to not touch the last 5% (or was
   it 10%) of NVRAM for exactly that reason.)
3. I'm convinced that "booting from file" (or via EFI shell) and running
   that 'efibootmgr -o 0000,0001,0002' from the initial comment would have
   restored the system, without ever touching TW rescue.
4. Can you tell how much NVRAM is in your system, and how much of that
   is occupied?  Is there anything eating up precious space?  I've seen
   issues like yours only on systems used for heavy testing with lots and
   lots of boot entries, or back then when that UEFI pstore-backend was
   active, or on buggy firmware (even from "big players")...
5. If you encounter the same issue again, e.g. after another grub2-update,
   could you please try remark 3 -- if that would fail, while efibootmgr
   from TW rescue succeeds...  but you actually did not *run* efibootmgr
   from TW rescue, did you?


You are receiving this mail because: