93936cbe14
Trace function graph after a page fault
2024-07-10 11:34:40 +02:00
6a63798308
The memtool fill mode also gets stuck
2024-07-10 11:13:05 +02:00
91080535f8
Add fill mode in memtool
2024-07-10 11:05:20 +02:00
6c8d6354bc
Add debug in memtool to see where it hangs
2024-07-09 22:12:55 +02:00
6b3af5b188
Reduce RAM to 1 GiB
2024-07-09 21:11:44 +02:00
813e7cae2f
Increase memtool max block size to 512 K
2024-07-09 20:58:55 +02:00
9ed8f812c8
Make our own init script
...
The problem with the Stage 1 script is that is spawning another tee
process to forward the output to the console, but this is not required.
So let's only keep the memtool running by creating our own init script.
2024-07-09 20:54:55 +02:00
c6726cce28
Reduce blocksize in memtool to 64K
...
Let's see if we can hang a bit later on.
2024-07-09 17:58:12 +02:00
4707a8b143
Simplify boot parameters and document options
2024-07-09 17:27:07 +02:00
b8f1ca672f
Run the memtool as first program
2024-07-09 17:25:48 +02:00
d488c0b3b7
Always run memtool at the start
2024-07-09 17:06:50 +02:00
5a88ed36b4
Managed to reproduce the crash consistently!
2024-07-09 15:38:52 +02:00
f617efdcac
Add memtool program to test the memory
2024-07-09 15:16:05 +02:00
82630f3eef
Enable CONFIG_BUG again
...
It doesn't seem to make any difference, and it would be good in case it
catches a bug.
2024-07-09 12:35:53 +02:00
dd6082e805
Add results of changing the CSR
...
It seems to arrive to systemd with all-in-order, but hangs there.
2024-07-09 12:26:00 +02:00
5b34b3b97b
Add csrtool to view and change CSR registers
2024-07-08 19:19:05 +02:00
bef5a6eac5
Disable CONFIG_BUG
...
It may be causing a loop on WARN_ON_ONCE() and page fults trying to
write to the console.
2024-07-08 18:27:29 +02:00
5f90528b51
Add bootrom support
2024-07-08 18:11:11 +02:00
66ec07a0cb
Switch to two uarts bitstream
2024-07-08 13:48:20 +02:00
1f0ac64631
Add bitstream to Nix
2024-07-08 13:46:37 +02:00
4d246ad00e
Enable secondary serial console
2024-07-08 10:44:51 +02:00
4641e0d9a0
Document hang missing hvc_remove trace point
2024-07-08 10:41:50 +02:00
6721e1e22c
Revert "Try to use openpiton based OpenSBI config"
...
This reverts commit 931244a355
2024-07-08 10:03:48 +02:00
1b4ab09c16
Select the second FPGA by default
2024-07-08 09:55:13 +02:00
931244a355
Try to use openpiton based OpenSBI config
...
The seem to be doing PLIC initialization based on the device tree, which
may be relevant as that is where we are defining the interruptions.
2024-07-08 09:49:34 +02:00
87b4c91813
Dump stack trace on hvc_remove
2024-07-08 08:55:50 +02:00
aaca0bb2e6
Reached stage2!
2024-07-08 08:49:21 +02:00
ef7a100c3f
Disable secondary console
2024-07-08 08:49:12 +02:00
6155c7e3f8
Try to fill cache details
2024-07-08 08:48:56 +02:00
a7c460b034
Use headings to allow hrefs
2024-07-05 17:05:28 +02:00
540780e508
Remove rvb dependency for now to avoid clang
2024-07-05 16:54:14 +02:00
d52691ca33
Add journal to the repository
2024-07-05 16:53:07 +02:00
254d1bd82b
flake.lock: Update
...
Flake lock file updates:
• Updated input 'nixpkgs':
'github:NixOS/nixpkgs/e4ad989506ec7d71f7302cc3067abd82730a4beb' (2023-11-19)
→ 'github:NixOS/nixpkgs/9f4128e00b0ae8ec65918efeba59db998750ead6' (2024-07-03)
2024-07-05 16:31:08 +02:00
b49471ba50
Update nixpkgs to nixos-unstable
2024-07-05 16:30:29 +02:00
b903eae7e5
Disable ftrace for now as it hangs the boot
2024-07-05 16:24:51 +02:00
7e6fdff142
Disable proactive compation
...
I see a lot of kcompactd0 entries in sched_switch log, so disable it for
now.
2024-07-05 15:57:18 +02:00
32f4d117db
Only enable sched_switch events
...
As sched_stat_runtime is flooding the console.
2024-07-05 15:37:50 +02:00
bc05dc2676
Mount debugfs first
2024-07-05 15:36:42 +02:00
934b67ac0b
Trace sched events from stage1
...
Events from the kthread that dumps the buffer to the console are
filtered, otherwise we make an infinite loop.
2024-07-05 15:36:42 +02:00
f509a0a517
Enable ftrace at boot
...
For now we only show init calls.
2024-07-05 15:36:37 +02:00
504c4aca3c
Make the counter silent by spawning another shell
2024-07-05 12:57:10 +02:00
48688594bf
Add a heartbeat counter to check the kernel
2024-07-05 12:45:32 +02:00
05898c5f85
Revert UART speed to 50MHz following vivado log
...
> UART Name g_UART IntfLabel m_axi_uart SyncClk {Label CLK0 Freq 50000000
> Name clk_i} BaseAddr 0x0 Number 2 AxiAddrWidth 0 AxiDataWidth 0 AxiIdWidth
> 0 AxiUserWidth 0 Mode normal IRQ uart_irq
2024-07-04 17:57:26 +02:00
7eeb16ecfb
Reduce frequency of UART clock to 25 MHz
2024-07-04 17:30:57 +02:00
d541462b4c
Stop in the stage 1 debug shell
2024-07-04 17:28:15 +02:00
9b37e2aed0
Use the hvc0 console for the kernel
2024-07-04 17:27:43 +02:00
eb7679f6a2
Prepare device tree to accomodate another UART
2024-07-04 17:22:43 +02:00
fcf4977a65
Extend the serial range from 0x100 to 0x1000
...
The AXI UART 16550 v2.0 from Xilinx only seem to require 32 bytes for
the registes, but let's reduce the changes with the original DTS.
2024-07-04 16:47:56 +02:00
d5b5cc3363
Change the order of interrupts-extended for PLIC
...
In the SiFive DTS they are in this order, not sure it that may cause any
difference.
2024-07-04 16:46:35 +02:00
471207c64a
Allow access to all memory from userspace
...
> If this option is disabled, you allow userspace (root) access to all
> of memory, including kernel and userspace memory
Currently is failing:
~ # cat /proc/iomem
40001000-400010ff : serial
60000000-7fffffff : Reserved
80000000-ffefffff : System RAM
80201000-81fa0b87 : Kernel image
80201000-80cb177f : Kernel code
81400000-819fffff : Kernel rodata
81c00000-81f18747 : Kernel data
81f19000-81fa0b87 : Kernel bss
100000000-1bfffffff : namespace0.0
~ # devmem 0x40001000
devmem: mmap: Operation not permitted
2024-07-04 12:45:41 +02:00
