Re: Xen Booting Problem on ARM Machine

[Sai Kiran Kumar Reddy Y <ysaikiran1997@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 4235 bytes --]

Hi,

Sorry about the delay. We have been trying to access the serial of the
machine. Tried with couple of JTAG connectors. There's still no debug
messages on the serial. Is there any other way of figuring this out?

On Wed, Sep 15, 2021, 7:02 AM Stefano Stabellini <sstabellini@kernel.org>
wrote:

> Something is off. When you enabled earlyprintk in Xen, you should see
> something like this at boot time:
> https://marc.info/?l=xen-devel&m=158829968025334
>
> All the Xen logs starting with "(XEN)" on the serial. Do you have access
> to the serial of the machine? Without it, it is going to be hard to
> debug.
>
>
> On Tue, 14 Sep 2021, Sai Kiran Kumar Reddy Y wrote:
> > In the folder "/var/log", there's a file called "xen", which is empty.
> As far as the boot logs are concerned, I don't see any debug
> > messages related to xen. I am attaching the log files, "kern.txt" and
> "boot.txt"
> >
> > On Tue, Sep 14, 2021 at 3:08 AM Stefano Stabellini <
> sstabellini@kernel.org> wrote:
> >       On Mon, 13 Sep 2021, Sai Kiran Kumar Reddy Y wrote:
> >       > On Fri, Sep 10, 2021 at 7:30 PM Julien Grall <julien@xen.org>
> wrote:
> >       >
> >       >
> >       >       On 08/09/2021 11:43, Sai Kiran wrote:
> >       >       > Hello,
> >       >
> >       >       Hi,
> >       >
> >       >       Thank you for the report. Moving the discussion back to
> xen-devel
> >       >       because this looks like a potential issue in the UEFI stub
> in Xen.
> >       >
> >       >       > I have Xen-4.15.0 on an ARM Machine, with Debian 11
> installed on it.
> >       >
> >       >       Would you be able to give more details on the Arm machine
> you are using?
> >       >       Also, are you using ACPI or DT to boot?
> >       >
> >       >
> >       >    Sai >> DT . ACPI configuration is disabled in Boot settings
>
> >       >
> >       >       >  I
> >       >       > am able to do “make world” and “make install”, after
> “./configure”, as
> >       >       > specified in README file. When I reboot the system, I
> get the following
> >       >       > message:
> >       >       >
> >       >       > Warning: All 128 bootinfo mem banks exhausted.
> >       >       >
> >       >       > Warning: All 128 bootinfo mem banks exhausted.
> >       >
> >       >       Hmmm... This means that you have more than 128 memory
> regions described
> >       >       in the EFI memory map. That's quite a lot.
> >       >
> >       >       Although, this should be harmless as it means Xen will not
> use the extra
> >       >       memory banks.
> >       >
> >       >       >
> >       >       > Cannot exit boot services: ErrCode: 0x8000000000000002
> >       >
> >       >       This means EFI_INVALID_PARAMETER. We have code to retry
> because AFAICT
> >       >       ExitBootServices() may sometime fails (I have CCed Jan may
> have more
> >       >       idea what's happening).
> >       >
> >       >       Would you be able to provide more details on the UEFI
> firmware you are
> >       >       using? Is it EDK2 or U-boot?
> >       >
> >       > Sai >>  EDK2
> >       >       Also, do you know if Linux is boot on the same system?
> >       >
> >       > Sai >> Yes
> >       >       However, AFAICT, the error message would not prevent Xen
> to continue
> >       >       booting. So you may get stuck later in the boot process.
> >       >
> >       >       My suggestion would be to enable earlyprintk for your
> platform. You can
> >       >       setup it up from the menuconfig in "Debugging Options".
> >       >
> >       > Sai >> Yes, I have enabled earlyprintk.
> >       > I tried changing NR_MEM_BANKS(in xen/include/asm-arm/setup.h)
> value to 256, from 128. The error message is no longer seen,
> >       but the device
> >       > is stuck in the boot process.
> >
> >       Could you please post the boot logs now that you enabled
> earlyprintk?
> >       Ideally not a camera picture but a textual copy/paste from the
> target
> >       serial?
> >
> >       Earlyprintk is pretty verbose, we should be able to figure out
> where it
> >       gets stuck.
> >
> >
> >

[-- Attachment #2: Type: text/html, Size: 5841 bytes --]