Commit Graph

86 Commits

Author SHA1 Message Date
Tim Newsome 09bf86e31a Keep around cmderr for callers to inspect.
Use this to only change abstract register access behavior when cmderr
explicitly says the requested operation is unsupported.
2017-07-12 14:36:09 -07:00
Tim Newsome 856f70fe44 Try abstract register writes as well. 2017-07-12 14:13:31 -07:00
Tim Newsome f37e93bbc0 Try using abstract commands to read registers
This is the only way the spec guarantees that GPRs are accessible, and
depending on the implementation this might be the only way that CSRs are
accessible.

Also changed the debug code that parses out DMI fields to be simpler to
maintain (albeit a little slower).

riscv013_execute_debug_buffer() now automatically clears cmderr if the
command fails. That feels like the right behavior. (It does return the
error to its caller.)
2017-07-12 14:13:31 -07:00
Tim Newsome 4072fa493b Disable debugger-set triggers on connect
When first connecting to a target, have the debugger disable any
hardware triggers that are set by a previously connected debugger.
The 0.11 code already did this, but 0.13 did not.

To achieve this I decided to share the code to enumerate triggers
between 0.11 and 0.13, which required me to implement get_register() and
set_register() for 0.11, which made the whole change a lot larger than
you might have guessed.

Hopefully this sets us up to in the future share the code to set/remove
triggers as well.
2017-07-10 10:26:24 -07:00
Tim Newsome 31e5b53a46 Merge pull request #74 from riscv/build32
Fix 32-bit build errors.
2017-07-06 13:41:47 -07:00
Palmer Dabbelt 3cff4213a4 Merge pull request #69 from riscv/multi-gdb
Fix the multi-GDB mode bugs
2017-07-03 13:18:06 -07:00
Tim Newsome 450307b66f Fix 32-bit build errors.
I only compiled the source. Didn't have the tooling installed to link.
Hopefully that's good enough.
Fixes #71.
2017-07-03 12:17:07 -07:00
Tim Newsome f18fd83ac7 Fix trigger set/clear bug. 2017-07-03 11:52:35 -07:00
Dmitry Ryzhov 99a3673507 Fix comment about saving the temporary register in examine procedure. 2017-07-01 15:09:23 +03:00
Dmitry Ryzhov 7d451e00f5 Restore value of temporary register (s0) in examine OpenOCD procedure in case of core can not execute 64 bit instruction. 2017-06-30 19:15:58 +03:00
Palmer Dabbelt d77c4a953c Don't set breakpoints on disabled harts 2017-06-21 12:25:20 -07:00
Palmer Dabbelt 4bdb042224 Allow memory writes to proceed on all harts 2017-06-21 12:25:19 -07:00
Palmer Dabbelt a277416a39 Refactor examine, to avoid some assertions
Now that we're supporting non-RTOS multi-hart mode there's some more
assertions that you're running on the right hart.  Those assertions
aren't sane very early in examine, so I avoid them.
2017-06-21 12:25:19 -07:00
Palmer Dabbelt 788908fcf0 Factor out checking if harts should be used
Rather than having a bunch of "if rtos" stuff, I now just check "if
hart_enabled".  This makes some code paths cleaner, all of which were
buggy in the non-RTOS multi-hart mode.
2017-06-21 10:09:16 -07:00
Palmer Dabbelt 9f4cac5a38 Set current_hartid from coreid
This avoids a bunch of RTOS special cases.
2017-06-20 17:19:05 -07:00
Tim Newsome 9cd98058a0 Set hardware triggers on all harts.
Right now we're using "threads" to represent harts. gdb/OpenOCD assume
there's only one set of hardware breakpoints among all threads. Make it
so.
2017-06-20 13:10:35 -07:00
Tim Newsome ccdd26e3ef Comment curious code. 2017-06-20 11:32:42 -07:00
Tim Newsome 927f9d8873 Update list of "threads" when harts are discovered.
This ensures that "info threads" is accurate as soon as gdb connects.
Also print out number of triggers that is discovered in examine().
2017-06-20 11:32:42 -07:00
Tim Newsome 4d264b3579 Put early DEBUG notice of XLEN back. 2017-06-19 08:46:02 -07:00
Tim Newsome 6082f35a55 Update debug_defines. Clarify debug output.
Update debug_defines from the spec, commit 920ec9a690.
Decode dmstatus scans in the debug output.
2017-06-16 14:02:25 -07:00
Tim Newsome fd81f7fcac Fix comment. 2017-06-16 14:02:25 -07:00
Tim Newsome 851849a295 Tell the user about detected harts. 2017-06-16 14:02:25 -07:00
Tim Newsome ac2da40f74 Fix indentation to match OpenOCD style.
This change is just in the whitespace. There are no code changes.

See http://openocd.org/doc-release/doxygen/stylec.html
2017-06-15 12:44:50 -07:00
Tim Newsome 363a0a2bf2 Merge pull request #64 from riscv/release-fixes
Two fixes from the release branch
2017-06-15 12:43:46 -07:00
Tim Newsome 50a223ef9a Fix print statements to work with 64-bit addresses 2017-06-15 12:24:37 -07:00
Palmer Dabbelt 099a3020d2 Clear abstract errors from register_read_direct 2017-06-15 12:16:24 -07:00
Tim Newsome 64af052911 Fix the build.
Main change is to make riscv_addr_t be unsigned. The rest is mechanical
fixing of types, print statements, and a few signed/unsigned compares.

Smoketest indicates everything is working more or less as before.
2017-06-13 12:33:01 -07:00
Tim Newsome 6be600318c Fix dmi_read() indentation; remove \n in LOG_ERROR 2017-06-08 12:31:08 -07:00
Megan Wachs c3b344d1c0 riscv: Move the initialization of the field inside the structure for consistency 2017-06-07 21:06:33 -07:00
Megan Wachs 459b39ec67 riscv: v13 -- dmi_write must still check for the OP result 2017-06-07 21:06:33 -07:00
Palmer Dabbelt c431c0eb25 Check for abstractcs.busy, not just CMDERR_BUSY
This fixes a race condition when reading/writing memory.
2017-05-15 17:40:28 -07:00
Palmer Dabbelt a8cf04b839 Go back to 32-word read/write buffers
The larger buffers are really slow on Spike.
2017-05-15 16:57:25 -07:00
Palmer Dabbelt e31761df64 Don't re-read registers after they're written
This was just a sanity check.
2017-05-15 16:57:08 -07:00
Palmer Dabbelt 9d308db2bc Print out the actual CSR that's read 2017-05-15 16:56:50 -07:00
Palmer Dabbelt 8252b9d36c Build fixes 2017-05-15 13:39:58 -07:00
Megan Wachs af6e04d5c0 riscv: Remove some compile warnings 2017-05-15 13:36:05 -07:00
Palmer Dabbelt 563f6acc3c Allow all harts to be reset 2017-05-09 13:33:20 -07:00
Megan Wachs 95a2eb157a riscv-013: more consistent parens 2017-05-01 09:42:11 -07:00
Megan Wachs 458bb20699 riscv-013: Correct sign extension of address on read_memory for lower bits as well 2017-05-01 09:39:59 -07:00
Megan Wachs 8462750357 riscv-013: Correct sign extension of address on read_memory 2017-05-01 09:37:48 -07:00
Megan Wachs ad1cf13ef4 Correct debugging print in read_memory 2017-05-01 08:35:10 -07:00
Palmer Dabbelt 16de5044d4 Fix an assertion when reading from 0 2017-05-01 08:33:01 -07:00
Palmer Dabbelt 1ec607c726 Clean up unused read_memory code 2017-04-27 12:56:01 -07:00
Palmer Dabbelt 8dea2908b7 Add 64-bit and multihart support
This is a major rewrite of the RISC-V v0.13 OpenOCD port.  This
shouldn't have any meaningful effect on the v0.11 support, but it does
add generic versions of many functions that will allow me to later
refactor the v0.11 support so it's easier to maintain both ports.  This
started as an emergency feature branch and went on for a long time, so
it's all been squashed down into one commit so there isn't a big set of
broken commits lying around.  The changes are:

 * You can pass "-rtos riscv" to the target in OpenOCD's configuration
   file, which enables multi-hart mode.  This uses OpenOCD's RTOS
   support to control all the harts from the debug module using commands
   like "info threads" in GDB.  This support is still expermental.

 * There is support for RV64I, but due to OpenOCD limitations we only
   support 32-bit physical addresses.  I hope to remedy this by rebasing
   onto the latest OpenOCD release, which I've heard should fix this.

 * This matches the latest draft version of the RISC-V debug spec, as of
   April 26th.  This version fixes a number of spec bugs and should be
   close to the final debug spec.
2017-04-26 09:10:49 -07:00
Megan Wachs 3dc066382b Properly consider 'reset halt' and do halt or resume as needed 2017-04-10 12:03:15 -07:00
Megan Wachs b04c6cb3ed riscv: Implement the assert/deassert reset functions for v13 2017-04-10 12:03:15 -07:00
Megan Wachs 14e26040b8 riscv: move value read to after autoexec is cleared. 2017-04-04 16:33:17 -07:00
Megan Wachs 9c1f6ea28b riscv: Correct the autoexec in read_mem 2017-04-04 16:22:55 -07:00
Megan Wachs eb90a5e05e riscv: Use write-1-to-clear for CMDERR, not write 0 to clear. 2017-03-30 14:27:28 -07:00
Palmer Dabbelt 03fa9d22c4 Don't set abstractauto at the start 2017-03-23 14:50:40 -07:00