[U-Boot] coreboot and u-boot integration x86: "No tick base available"
Hi.
I'm trying to use u-boot as a payload to coreboot. Problem is that when coreboot starts u-boot, it fails with panic("No tick base available");
When looking at the backtrace this gives a recursive error, as panic calls __udelay and get_ticks, which again panics. I heard this was because u-boot overwrote the memory location of coreboot, and that there have been some patches going around that fixes this issue, but have not made it upstream. As far as I can tell, chromebook v2 uses these patches to make their stuff boot. Can anyone point me in the right direction for those patches? I have tried finding them myself, but can't seem to find them.
Regards - Martin
Hi Martin,
On 30 May 2014 04:33, Martin Ertsås martiert@gmail.com wrote:
Hi.
I'm trying to use u-boot as a payload to coreboot. Problem is that when coreboot starts u-boot, it fails with panic("No tick base available");
When looking at the backtrace this gives a recursive error, as panic calls __udelay and get_ticks, which again panics. I heard this was because u-boot overwrote the memory location of coreboot, and that there have been some patches going around that fixes this issue, but have not made it upstream. As far as I can tell, chromebook v2 uses these patches to make their stuff boot. Can anyone point me in the right direction for those patches? I have tried finding them myself, but can't seem to find them.
It probably means that Coreboot is not passing its timing data to U-Boot. You need to enable a timestamp option in Coreboot to do this.
You could patch it to remove this panic and just use 0 in this case.
Regards, Simon
On 06/01/14 18:42, Simon Glass wrote:
Hi Martin,
On 30 May 2014 04:33, Martin Ertsås martiert@gmail.com wrote:
Hi.
I'm trying to use u-boot as a payload to coreboot. Problem is that when coreboot starts u-boot, it fails with panic("No tick base available");
When looking at the backtrace this gives a recursive error, as panic calls __udelay and get_ticks, which again panics. I heard this was because u-boot overwrote the memory location of coreboot, and that there have been some patches going around that fixes this issue, but have not made it upstream. As far as I can tell, chromebook v2 uses these patches to make their stuff boot. Can anyone point me in the right direction for those patches? I have tried finding them myself, but can't seem to find them.
It probably means that Coreboot is not passing its timing data to U-Boot. You need to enable a timestamp option in Coreboot to do this.
You could patch it to remove this panic and just use 0 in this case.
Regards, Simon
Thanks. I'll look into the timestamp option. So using 0 as the tick value should work?
- Martin
Hi Martin,
On 1 June 2014 23:59, Martin Ertsås martiert@gmail.com wrote:
On 06/01/14 18:42, Simon Glass wrote:
Hi Martin,
On 30 May 2014 04:33, Martin Ertsås martiert@gmail.com wrote:
Hi.
I'm trying to use u-boot as a payload to coreboot. Problem is that when coreboot starts u-boot, it fails with panic("No tick base available");
When looking at the backtrace this gives a recursive error, as panic calls __udelay and get_ticks, which again panics. I heard this was because u-boot overwrote the memory location of coreboot, and that there have been some patches going around that fixes this issue, but have not made it upstream. As far as I can tell, chromebook v2 uses these patches to make their stuff boot. Can anyone point me in the right direction for those patches? I have tried finding them myself, but can't seem to find them.
It probably means that Coreboot is not passing its timing data to U-Boot. You need to enable a timestamp option in Coreboot to do this.
You could patch it to remove this panic and just use 0 in this case.
Regards, Simon
Thanks. I'll look into the timestamp option. So using 0 as the tick value should work?
Yes, although you won't get boot timing from Coreboot then. One of the engineers favoured a panic() to avoid accidentally dropping the function from Coreboot's build. Perhaps it could be changed to be a default in Coreboot? This is the second time the issue has come up in U-Boot.
Also I'd be happy with printing a warning in this case if you want to do a patch.
Regards,
On 06/03/14 03:43, Simon Glass wrote:
Hi Martin,
On 1 June 2014 23:59, Martin Ertsås martiert@gmail.com wrote:
On 06/01/14 18:42, Simon Glass wrote:
Hi Martin,
On 30 May 2014 04:33, Martin Ertsås martiert@gmail.com wrote:
Hi.
I'm trying to use u-boot as a payload to coreboot. Problem is that when coreboot starts u-boot, it fails with panic("No tick base available");
When looking at the backtrace this gives a recursive error, as panic calls __udelay and get_ticks, which again panics. I heard this was because u-boot overwrote the memory location of coreboot, and that there have been some patches going around that fixes this issue, but have not made it upstream. As far as I can tell, chromebook v2 uses these patches to make their stuff boot. Can anyone point me in the right direction for those patches? I have tried finding them myself, but can't seem to find them.
It probably means that Coreboot is not passing its timing data to U-Boot. You need to enable a timestamp option in Coreboot to do this.
You could patch it to remove this panic and just use 0 in this case.
Regards, Simon
Thanks. I'll look into the timestamp option. So using 0 as the tick value should work?
Yes, although you won't get boot timing from Coreboot then. One of the engineers favoured a panic() to avoid accidentally dropping the function from Coreboot's build. Perhaps it could be changed to be a default in Coreboot? This is the second time the issue has come up in U-Boot.
Also I'd be happy with printing a warning in this case if you want to do a patch.
Regards,
Ok. Looked into enabeling it in coreboot though, but couldn't find a way to do it with a qemu machine. Guess I'll have to nag the coreboot developers a bit :)
In one way it was kind of frustrating, but I do see the reason for having it as a panic though. You have probably done something wrong if you forgot to enable ticks. I don't feel like I know u-boot well enough to make a decision if it should be a panic or a warning though. If you would preffer a warning I can make the patch.
- Martin
Hi Martin,
On 3 June 2014 00:22, Martin Ertsås martiert@gmail.com wrote:
On 06/03/14 03:43, Simon Glass wrote:
Hi Martin,
On 1 June 2014 23:59, Martin Ertsås martiert@gmail.com wrote:
On 06/01/14 18:42, Simon Glass wrote:
Hi Martin,
On 30 May 2014 04:33, Martin Ertsås martiert@gmail.com wrote:
Hi.
I'm trying to use u-boot as a payload to coreboot. Problem is that when coreboot starts u-boot, it fails with panic("No tick base available");
When looking at the backtrace this gives a recursive error, as panic calls __udelay and get_ticks, which again panics. I heard this was because u-boot overwrote the memory location of coreboot, and that there have been some patches going around that fixes this issue, but have not made it upstream. As far as I can tell, chromebook v2 uses these patches to make their stuff boot. Can anyone point me in the right direction for those patches? I have tried finding them myself, but can't seem to find them.
It probably means that Coreboot is not passing its timing data to U-Boot. You need to enable a timestamp option in Coreboot to do this.
You could patch it to remove this panic and just use 0 in this case.
Regards, Simon
Thanks. I'll look into the timestamp option. So using 0 as the tick value should work?
Yes, although you won't get boot timing from Coreboot then. One of the engineers favoured a panic() to avoid accidentally dropping the function from Coreboot's build. Perhaps it could be changed to be a default in Coreboot? This is the second time the issue has come up in U-Boot.
Also I'd be happy with printing a warning in this case if you want to do a patch.
Regards,
Ok. Looked into enabeling it in coreboot though, but couldn't find a way to do it with a qemu machine. Guess I'll have to nag the coreboot developers a bit :)
In one way it was kind of frustrating, but I do see the reason for having it as a panic though. You have probably done something wrong if you forgot to enable ticks. I don't feel like I know u-boot well enough to make a decision if it should be a panic or a warning though. If you would preffer a warning I can make the patch.
I don't have a strong opinion, so if you are happy with it as it is, that is fine with me. But if not, please send a patch.
Regards, Simon
participants (2)
-
Martin Ertsås -
Simon Glass