Re: Replacing existing kernel with new on qemuriscv64

On Mon, Jul 20, 2020 at 10:31 PM Pankaj Vinadrao Joshi
<Pankaj.VJ@exaleapsemi.com> wrote:
>
> Hi,
> i will explain what i mean is, i am using sifive's u540 hardware for which we have built our own custom kernel using openembedded-core lets say 5.5.6 and now i want to build the other kernel lets say 5.7, natively

Natively for which platform? You want to do build a new kernel on the
hardware and then boot into that? The kernel's are kept in a boot
partition, you should just have to copy them to that location.

> and replace the old kernel with new one,and for this i really did not understood where and what exactly i should modify to boot from the new kernel??

Yep, just replace the old one.

>
> The steps what i followed are following
>
> 1) i have cloned the kernel source
> 2)make menuconfig
> 3)make -j4
> 4)make modules_install
> 5make install

make install is probably not going to understand what to do. I think
you will need to manually do this. I can't remember where the
partition is, but it should be easy to find.

Alistair

>
> Hope now its clear what i was trying to ask for??
>
> Thanks & regards
> Pankaj
>
>
>
> ________________________________
> From: Alistair Francis <alistair23@gmail.com>
> Sent: Tuesday, July 21, 2020 12:10 AM
> To: Pankaj Vinadrao Joshi <Pankaj.VJ@exaleapsemi.com>
> Cc: qemu-devel@nongnu.org <qemu-devel@nongnu.org>
> Subject: Re: Replacing existing kernel with new on qemuriscv64
>
> On Mon, Jul 20, 2020 at 11:19 AM Pankaj Vinadrao Joshi
> <Pankaj.VJ@exaleapsemi.com> wrote:
> >
> > Hi Alistair Francis,
> >
> > Thanks for your response.I have same concern for other riscv hardware with custom kernel will same be applicable for it?If no from where i should change my kernel image since there also i am not able to find grub and the clarification given on web are seems to be very specific to x86 arch where i should look in case of riscv ??
>
> Hello,
>
> I'm not really sure what you are asking.
>
> The way to update the kernel will depend on how you are booting it and
> where it is stored. GRUB supports RISC-V, but as there is no UEFI
> support in the kernel (yet) it probably isn't the best boot method.
> Most hardware will probably use u-boot instead.
>
> Alistair
>
> >
> > Thanks
> >
> > ________________________________
> > From: Alistair Francis <alistair23@gmail.com>
> > Sent: Monday, July 20, 2020 10:29 PM
> > To: Pankaj Vinadrao Joshi <Pankaj.VJ@exaleapsemi.com>
> > Cc: qemu-devel@nongnu.org <qemu-devel@nongnu.org>
> > Subject: Re: Replacing existing kernel with new on qemuriscv64
> >
> > On Mon, Jul 20, 2020 at 2:46 AM Pankaj Vinadrao Joshi
> > <Pankaj.VJ@exaleapsemi.com> wrote:
> > >
> > > Hi ,
> > > I am trying to replace my existing kernel image which is 5.5.6 to 5.5.7 on qemuriscv64 i would like to mention i have built my qemu image with openembedded-core.
> > >
> > > i have tried to build the kernel 5.5.7 natively,i was able to build the kernel successfully and in my /boot folder i am able to see vmlinux-5.5.7 but now i want to install this image..i have tried to do it by make install i didnt got any error with it but when i rebooted my system i am getting
> > >
> > > i have followed the following steps to build kernel
> > > $root@qemuriscv64-exaleapsemi-r2:/usr/src/kernel# make menuconfig arch=riscv64
> > > $root@qemuriscv64-exaleapsemi-r2:/usr/src/kernel# make -j4
> > > $root@qemuriscv64-exaleapsemi-r2:/usr/src/kernel# make modules_install
> > > $root@qemuriscv64-exaleapsemi-r2:/usr/src/kernel# make install
> > > sh ./arch/riscv/boot/install.sh 5.5.7 \
> > > arch/riscv/boot/Image System.map "/boot"
> > > Installing normal kernel
> > >
> > > $reboot
> > > 13024.451157] printk: systemd-shutdow: 37 output lines suppressed due to ratelimiting
> > > [13024.527282] systemd-shutdown[1]: Syncing filesystems and block devices.
> > > [13024.668538] systemd-shutdown[1]: Sending SIGTERM to remaining processes...
> > > [13024.719496] systemd-journald[87]: Received SIGTERM from PID 1 (systemd-shutdow).
> > > [13024.769405] systemd-shutdown[1]: Sending SIGKILL to remaining processes...
> > > [13024.834318] systemd-shutdown[1]: Unmounting file systems.
> > > [13024.868285] [13413]: Remounting '/' read-only in with options '(null)'.
> > > [13025.228499] EXT4-fs (vda): re-mounted. Opts: (null)
> > > [13025.358123] systemd-shutdown[1]: All filesystems unmounted.
> > > [13025.358576] systemd-shutdown[1]: Deactivating swaps.
> > > [13025.361264] systemd-shutdown[1]: All swaps deactivated.
> > > [13025.361727] systemd-shutdown[1]: Detaching loop devices.
> > > [13025.420602] systemd-shutdown[1]: All loop devices detached.
> > > [13025.421071] systemd-shutdown[1]: Detaching DM devices.
> > > [13025.595237] reboot: Restarting system
> > > [13046.599875] rcu: INFO: rcu_sched detected stalls on CPUs/tasks:
> > > [13046.600329] (detected by 0, t=5252 jiffies, g=2837145, q=17)
> > > [13046.600763] rcu: All QSes seen, last rcu_sched kthread activity 5252 (4298153942-4298148690), jiffies_till_next_fqs=1, root ->qsmask 0x0
> > > [13046.601503] systemd-shutdow R running task 0 1 0 0x00000008
> > > [13046.602053] Call Trace:
> > > [13046.602562] [<ffffffe00006d3fc>] walk_stackframe+0x0/0xa4
> > > [13046.602890] [<ffffffe00006d5e2>] show_stack+0x2a/0x34
> > > [13046.603156] [<ffffffe000091094>] sched_show_task+0x158/0x174
> > > [13046.603442] [<ffffffe0000c4dde>] rcu_sched_clock_irq+0x6b2/0x6ec
> > > [13046.603740] [<ffffffe0000ccadc>] update_process_times+0x1e/0x44
> > > [13046.604009] [<ffffffe0000d79dc>] tick_sched_handle.isra.16+0x30/0x3e
> > > [13046.604331] [<ffffffe0000d7a38>] tick_sched_timer+0x4e/0x94
> > > [13046.604620] [<ffffffe0000cd44a>] __hrtimer_run_queues+0x10c/0x2de
> > > [13046.604928] [<ffffffe0000cdc96>] hrtimer_interrupt+0xcc/0x1d6
> > > [13046.605236] [<ffffffe0005d3f7c>] riscv_timer_interrupt+0x32/0x3a
> > > [13046.605580] [<ffffffe0007889ce>] do_IRQ+0xae/0xba
> > > [13046.605837] [<ffffffe00006bd3c>] ret_from_exception+0x0/0xc
> > > [13046.606127] [<ffffffe00006ca58>] machine_restart+0xc/0xe
> > > [13046.606508] rcu: rcu_sched kthread starved for 5252 jiffies! g2837145 f0x2 RCU_GP_WAIT_FQS(5) ->state=0x0 ->cpu=0
> > > [13046.606990] rcu: RCU grace-period kthread stack dump:
> > > [13046.607257] rcu_sched R running task 0 10 2 0x00000000
> > > [13046.607610] Call Trace:
> > > [13046.607774] [<ffffffe000784258>] __schedule+0x1a8/0x482
> > > [13046.608044] [<ffffffe000784592>] schedule+0x60/0xda
> > > [13046.608304] [<ffffffe000787a56>] schedule_timeout+0x15c/0x2b2
> > > [13046.608602] [<ffffffe0000c2ec8>] rcu_gp_kthread+0x4cc/0x7ec
> > > [13046.608889] [<ffffffe00008b0f2>] kthread+0xf0/0x102
> > > [13046.609162] [<ffffffe00006bd3c>] ret_from_exception+0x0/0xc
> > >
> > > ..... so on getting continuously......
> > >
> > >
> > > Can someone help me out how i can replace my kernel to 5.5.7 since i am not able to find grub or grub2 on my riscv platform?? i have googled for the solution but most of them seems to be x86 specific where they are trying to change the grub??Where i should modify for riscv to get my kernel up????
> >
> > You won't be able to install a new kernel from the QEMU guest. You
> > will need to use OpenEmbedded to build a new kernel and boot that.
> >
> > Alistair
> >
> > >
> > >
> > > Thanks
> > [EXT]
> [EXT]
[EXT]

From:	Pankaj Vinadrao Joshi
Subject:	Re: Replacing existing kernel with new on qemuriscv64
Date:	Thu, 23 Jul 2020 09:55:46 +0000