Bug 680542
Summary: | better error message when out of memory for initrd.img | ||
---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Charles R. Anderson <cra> |
Component: | kernel | Assignee: | Kernel Maintainer List <kernel-maint> |
Status: | CLOSED WONTFIX | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
Severity: | medium | Docs Contact: | |
Priority: | unspecified | ||
Version: | 16 | CC: | awilliam, bojan, gansalmon, gholms, itamar, jonathan, jones.peter.busi, kernel-maint, kvolny, madhu.chinakonda, robatino, stephent98, wwoods |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | RejectedBlocker | ||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2011-11-02 17:09:01 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: | |||
Attachments: |
Description
Charles R. Anderson
2011-02-25 20:49:07 UTC
Created attachment 481420 [details]
Add a user-friendly notifier to panic messages
This adds a user-friendly face to panic messages, akin to the SPARC HHGTG guy, or the PA-RISC cow.
Created attachment 491094 [details] screenshot showing kernel messages after out-of-memory panic This is what users will see if they do not have enough memory, instead of an informative error message. QEMU shows 100% CPU utilization at this point. $ qemu-kvm -m 512 Fedora-15-Beta-i386-netinst.iso F15-Beta-RC2 See also: Bug 682555 - Net-install dvd gets a kernel panic on install I have to power cycle my laptop after the kernel panics. It would be helpful if I could press ctrl-alt-del to restart. The kernel could suggest doing that. (In reply to comment #1) > Created attachment 481420 [details] > Add a user-friendly notifier to panic messages > > This adds a user-friendly face to panic messages, akin to the SPARC HHGTG guy, > or the PA-RISC cow. Perfect! What could possibly be more user-friendly? Created attachment 491334 [details]
Ease the user's concerns
This (untested) patch should provide more reasonable feedback to the user.
Created attachment 491336 [details]
Fixed patch
...although we should really make sure the machine stops running, too.
Discussed at 2011-04-15 blocker review meeting. We reject this as a blocker on the basis there is no practical code fix anyone can think of (though we whole-heartedly endorse mjg59's patch). We definitely support trying to ensure this is reasonably documented in the release notes, and the system requirements are updated to reflect reality. -- Fedora Bugzappers volunteer triage team https://fedoraproject.org/wiki/BugZappers The real fix would be to resurrect stage2.img. I think it was a mistake to drop it, considering the impact on RAM requirements. There are plans to reduce memory requirements for F16: Bug 682555, Comment 13. The patch looks fine to my less-than-expert eye: 1. Would the user be able to restart with ctrl-alt-del? 2. Can the strings be localized? 3. Could the graphic be made configurable? (In reply to comment #9) > There are plans to reduce memory requirements for F16: Bug 682555, Comment 13. Correct. Bug 682555 is the place to discuss that stuff. Not that there's much that needs discussing; splitting the boot image back into two parts (initramfs + runtime) is already the plan for F16. > 2. Can the strings be localized? Nothing else in the kernel is localized. There's nowhere to store the localized strings. The kernel hasn't even finished initializing yet - there's no storage, no disk drivers, no initrd, nothing. But the phrase "MORE RAM" should be pretty easy to suss out. > 3. Could the graphic be made configurable? Absolutely not. Why would you want to? Relax, friend, and bask in the glory of the Hot Dog. (In reply to comment #10) > (In reply to comment #9) ... > > 3. Could the graphic be made configurable? > > Absolutely not. Why would you want to? Relax, friend, and bask in the glory of > the Hot Dog. <g> Straight Man: http://en.wikipedia.org/wiki/Double_act Bud Abbott: http://en.wikipedia.org/wiki/Bud_Abbott Created attachment 498958 [details]
screenshot after F15-RC3 kernel out of memory while unpacking rootfs image
The F15-RC3 kernel appears to have disabled the call trace when out of memory while unpacking rootfs image. Now a message like this can be read at the bottom of the screen:
[ 0.171631] Trying to unpack rootfs image as initramfs...
The kernel appears to have hung, but qemu is consuming 100% CPU.
Tested with:
$ qemu-kvm -m 512 -cdrom F15-RC3/Fedora-15-x86_64-DVD.iso
I tried with -m 1024 and still got a hang. Weird... (In reply to comment #13) > I tried with -m 1024 and still got a hang. Weird... With a 3.2GHz processor, the unpacking takes 10 seconds. (In reply to comment #14) > (In reply to comment #13) > > I tried with -m 1024 and still got a hang. Weird... > > With a 3.2GHz processor, the unpacking takes 10 seconds. You are right. I have a much slower machine and was just being impatient. (In reply to comment #15) > (In reply to comment #14) > > (In reply to comment #13) > > > I tried with -m 1024 and still got a hang. Weird... > > > > With a 3.2GHz processor, the unpacking takes 10 seconds. > > You are right. I have a much slower machine and was just being impatient. Even 10 seconds is too long to wait For the kernel's user interface To say something The initramfs is 90MB of xz-compressed data, and 365MB uncompressed. Uncompressing that data takes a long time. In F16 we should only be uncompressing something like 4MB at boot time, so this won't be a problem. But this is pretty far from the original purpose of this bug, so let's try to get back to that. *** Bug 715572 has been marked as a duplicate of this bug. *** We clearly aren't going to get much benefit from this in F15. Moving to F16 so we can see if we want to throw the almighty Hot Dog OOM patch in there. F16 is about to ship. We didn't include the Hot Dog patch there either. The memory usage of the initramfs should actually be less now in F16 and further into F17. I'm going to close this out as WONTFIX as the situation should be less frequent now, and the patch doesn't really carry a whole lot of benefit. |