slow boot with 7fef431be9c9 ("mm/page_alloc: place pages to tail in __free_pages_core()")

[Deucher, Alexander]

[AMD Public Use]

> -----Original Message-----
> From: David Hildenbrand <david at redhat.com>
> Sent: Friday, March 12, 2021 10:48 AM
> To: Deucher, Alexander <Alexander.Deucher at amd.com>; linux-
> kernel at vger.kernel.org; amd-gfx list <amd-gfx at lists.freedesktop.org>;
> Andrew Morton <akpm at linux-foundation.org>; Liang, Liang (Leo)
> <Liang.Liang at amd.com>
> Cc: Huang, Ray <Ray.Huang at amd.com>; Koenig, Christian
> <Christian.Koenig at amd.com>; Mike Rapoport <rppt at linux.ibm.com>;
> Rafael J. Wysocki <rafael at kernel.org>; George Kennedy
> <george.kennedy at oracle.com>
> Subject: Re: slow boot with 7fef431be9c9 ("mm/page_alloc: place pages to
> tail in __free_pages_core()")
> 
> > 8G (with some carve out for the integrated GPU).
> > [    0.044181] Memory: 6858688K/7200304K available (14345K kernel code,
> 9659K rwdata, 4980K rodata, 2484K init, 12292K bss, 341360K reserved, 0K
> cma-reserved)
> >
> > Nothing particularly special about these systems that I am aware of.  I'll see
> if we can repro this issue on any other platforms, but so far, not one has
> noticed any problems.
> >
> >>
> >> Increasing the boot time from a few seconds to 2-3 minutes does not
> >> smell like some corner case cache effects we might be hitting in this
> >> particular instance - there have been minor reports that it either
> >> slightly increased or slightly decreases initial system performance, but that
> was about it.
> >>
> >> Either, yet another latent BUG (but why? why should memory access
> >> suddenly be that slow? I could only guess that we are now making
> >> sooner use of very slow memory), or there is really something else weird
> going on.
> >
> > Looks like pretty much everything is slower based on the timestamps in the
> dmesg output.  There is a big jump here:
> 
> If we're really dealing with some specific slow memory regions and that
> memory gets allocated for something that gets used regularly, then we might
> get a general slowdown. Hard to identify, though :)
> 
> >
> >> [    3.758596] ACPI: Added _OSI(Linux-Lenovo-NV-HDMI-Audio)
> >> [    3.759372] ACPI: Added _OSI(Linux-HPI-Hybrid-Graphics)
> >> [   16.177983] ACPI: 13 ACPI AML tables successfully acquired and loaded
> >> [   17.099316] ACPI: [Firmware Bug]: BIOS _OSI(Linux) query ignored
> >> [   18.969959] ACPI: EC: EC started
> >
> > And here:
> >
> >> [   36.566608] PCI: CLS 64 bytes, default 64
> >> [   36.575383] Trying to unpack rootfs image as initramfs...
> >> [   44.594348] Initramfs unpacking failed: Decoding failed
> >> [   44.765141] Freeing initrd memory: 46348K
> >
> > Also seeing soft lockups:
> >> [  124.588634] watchdog: BUG: soft lockup - CPU#1 stuck for 23s!
> >> [swapper/1:0]
> 
> Yes, I noticed that -- there is a heavy slowdown somewhere.
> 
> As that patch is v5.10 already (and we're close to v5.12) I assume something
> is particularly weird about the platform you are running on - because this is
> the first time I see a report like that.

Well, this platform is not yet widely available outside of AMD so it's not likely to have been seen by anyone else, but there is nothing special about it compared to any other AMD platforms beyond that that I am aware of.

Alex