M-Labs
/
artiq-zynq
10
1
Fork 19
Code Issues 14 Pull Requests 1 Releases Wiki Activity

set exception vector table addr to 0x100000 and add FIQ dummy handler #288

Merged
sb10q merged 1 commits from morgan/artiq-zynq:vector_fix into master 2024-08-17 17:37:23 +08:00
Conversation 6 Commits 1 Files Changed 3 +20 -2
morgan commented 2024-03-07 12:56:17 +08:00
Owner
Copy Link

Prerequisite

Summary

  • add dummy fiq handler to fix the compilation error after flake update (see zynq-rs/110)
  • use set_vector_table to configure the base address of the vector table to 0x100000 (the start of SDRAM)

Testing

  • Standlone with Urukul: pass sinara tester
  • Master -> Satellite with Urukul: pass sinara tester
## Prerequisite - flake update for [zynq-rs/#112](https://git.m-labs.hk/M-Labs/zynq-rs/pulls/112) and [zynq-rs/#111](https://git.m-labs.hk/M-Labs/zynq-rs/pulls/111) ## Summary - add dummy fiq handler to fix the compilation error after flake update (see [zynq-rs/110](https://git.m-labs.hk/M-Labs/zynq-rs/pulls/110)) - use `set_vector_table` to configure the base address of the vector table to `0x100000` (the start of SDRAM) - fix #287 ## Testing - Standlone with Urukul: pass sinara tester - Master -> Satellite with Urukul: pass sinara tester
sb10q commented 2024-03-07 13:10:09 +08:00
Owner
Copy Link

add dummy fiq handler to fix the compilation error after flake update (see zynq-rs/110)

Why can't it use the default handler in zynq-rs like it does for others?

> add dummy fiq handler to fix the compilation error after flake update (see zynq-rs/110) Why can't it use the default handler in zynq-rs like it does for others?
morgan force-pushed vector_fix from eee45ff828 to 9763a30ed2 2024-03-07 13:27:12 +08:00 Compare
sb10q reviewed 2024-03-07 13:27:46 +08:00
src/runtime/src/main.rs Outdated
@ -77,6 +77,7 @@ static mut LOG_BUFFER: [u8; 1 << 17] = [0; 1 << 17];
#[no_mangle]
pub fn main_core0() {
exception_vectors::set_vector_table(0x100000);
sb10q commented 2024-03-07 13:27:46 +08:00
Owner
Copy Link

Where is this number coming from?

Where is this number coming from?
sb10q commented 2024-03-07 13:28:02 +08:00
Owner
Copy Link

Maybe obtain from linker?

Maybe obtain from linker?
👍 1
morgan commented 2024-03-07 13:29:23 +08:00
Author
Owner
Copy Link

From the link.x, I also checked it with .elf dump


runtime.elf:     file format elf32-littlearm


Disassembly of section .text:

00100000 <exception_vector>:
  100000:	ea000036 	b	1000e0 <Reset>
  100004:	ea000045 	b	100120 <UndefinedInstruction>
  100008:	ea000054 	b	100160 <SoftwareInterrupt>
  10000c:	ea000063 	b	1001a0 <PrefetchAbort>
  100010:	ea000072 	b	1001e0 <DataAbort>
  100014:	ea000081 	b	100220 <ReservedException>
  100018:	ea000010 	b	100060 <IRQ>
  10001c:	eaffffff 	b	100020 <FIQ>
From the [link.x](https://git.m-labs.hk/M-Labs/artiq-zynq/src/commit/cf0b83c3f968833d410060a9ad981ad822c3690a/src/libbuild_zynq/link.x#L5), I also checked it with .elf dump ``` runtime.elf: file format elf32-littlearm Disassembly of section .text: 00100000 <exception_vector>: 100000: ea000036 b 1000e0 <Reset> 100004: ea000045 b 100120 <UndefinedInstruction> 100008: ea000054 b 100160 <SoftwareInterrupt> 10000c: ea000063 b 1001a0 <PrefetchAbort> 100010: ea000072 b 1001e0 <DataAbort> 100014: ea000081 b 100220 <ReservedException> 100018: ea000010 b 100060 <IRQ> 10001c: eaffffff b 100020 <FIQ> ```
sb10q commented 2024-03-07 13:30:10 +08:00
Owner
Copy Link

So use __text_start then?

So use __text_start then?
morgan commented 2024-03-07 13:27:56 +08:00
Author
Owner
Copy Link

Force push to use the dummy_fiq_handler feature from zynq-rs

Force push to use the `dummy_fiq_handler` feature from zynq-rs
morgan force-pushed vector_fix from 9763a30ed2 to 4455f740d2 2024-03-07 15:38:57 +08:00 Compare
morgan commented 2024-03-07 15:40:45 +08:00
Author
Owner
Copy Link

Force push to use linker symbol. __exception_start is added and used instead of __text_start. As it's pointing at the start of .text (0x0)

[     0.035728s]  INFO(runtime): __exception_start = 0x100000
[     0.040936s]  INFO(runtime): __text_start = 0x0
[     0.045276s]  INFO(runtime): __text_end = 0x1afde8
Force push to use linker symbol. `__exception_start` is added and used instead of `__text_start`. As it's pointing at the start of .text (0x0) ```bash [ 0.035728s] INFO(runtime): __exception_start = 0x100000 [ 0.040936s] INFO(runtime): __text_start = 0x0 [ 0.045276s] INFO(runtime): __text_end = 0x1afde8 ```
sb10q commented 2024-03-07 15:41:54 +08:00
Owner
Copy Link

Hmm, the __text_start symbol looks wrong then. It should not be 0. Maybe it's not used anywhere and could simply be removed?

Hmm, the __text_start symbol looks wrong then. It should not be 0. Maybe it's not used anywhere and could simply be removed?
sb10q merged commit 4455f740d2 into master 2024-03-07 15:42:32 +08:00
morgan commented 2024-03-07 15:46:19 +08:00
Author
Owner
Copy Link

__text_start is used in the core1.rs file. Removing it or making it point to the right address may cause some unknown side effect

`__text_start` is used in the [core1.rs](https://git.m-labs.hk/M-Labs/artiq-zynq/src/commit/4455f740d2f8477ef94eff24cac2b68dd5fdfd57/src/libksupport/src/kernel/core1.rs#L215) file. Removing it or making it point to the right address may cause some unknown side effect
sb10q commented 2024-03-07 15:54:49 +08:00
Owner
Copy Link

If code works correctly with incorrect values, then the code is buggy and should be fixed.

If code works correctly with incorrect values, then the code is buggy and should be fixed.
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels   
priority:critical

   
priority:high

   
priority:low

   
priority:medium
No Label
priority:critical
priority:high
priority:low
priority:medium
Milestone
Clear milestone
No items
No Milestone
Assignees
Clear assignees
No Assignees
2 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: M-Labs/artiq-zynq#288
No description provided.
Delete Branch "morgan/artiq-zynq:vector_fix"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?