David Brownell [Fri, 9 Oct 2009 06:51:50 +0000 (23:51 -0700)]
make PXA255 targets enumerate sort-of-OK
Startup now mostly works, except that the initial target state
is "unknown" ... previously, it refused to even start.
Getting that far required fixing the ircapture value (which
can never have been correct!) and the default JTAG clock rate,
then providing custom reset script.
The "reset" command is still iffy. DCSR updates, and loading
the debug handler, report numerous DR/IR capture failures.
But once that's done, "poll" reports that the CPU is halted
(which it shouldn't be, this was "reset run"!), due to the
rather curious reason "target-not-halted".
Summary: you still can't debug these parts, but it's closer.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
David Brownell [Fri, 9 Oct 2009 06:51:50 +0000 (23:51 -0700)]
add overridable Tcl "init_reset"
This abstracts the "jtag arp_init-reset" call into a method
called from OpenOCD startup and reset processing.
Platforms which have different requirements for how such hard
resets must be performed can now override "init_reset" instead
of needing to rebuild custom hacked versions of the server.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
David Brownell [Thu, 8 Oct 2009 20:15:18 +0000 (13:15 -0700)]
initial builds of OSK5912 boards need srst_pulls_trst
This is clearly noted in the hardware spec (section 5.2.3); it
works around a chip erratum: "If the MPU_RESET signal is used,
it may cause the EMIFS bus to lock."
I seem to have a board with such an initial build. The chip
is labeled XOMAP. Presumably, parts without that "X" prefix
(eXperimental) resolve this.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
The issue was that nothing disabled background polling during calls
from the TCL shell to "jtag_reset 1 1". Fix by moving the existing
poll-disable mechanism to the JTAG layer where it belongs, and then
augmenting it to always pay attention to TRST and SRST.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
David Brownell [Thu, 8 Oct 2009 16:16:01 +0000 (09:16 -0700)]
Stop ignoring most scan chain validation errors
Among other things this causes startup errors to kick in the
fallback "reset harder" logic during server startup. Comments
are also updated a bit, explaining what the various error paths
signify (in at least my observation).
There's one class of validation error that we can still plausibly
ignore: when wrong IDCODE values are observed.
This change seems to have helped make an OMAP5912 behave much
more reliably. There's still some post-reset flakiness, but
it's unrelated to scan verification.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
dbrownell [Wed, 7 Oct 2009 19:02:21 +0000 (19:02 +0000)]
Force sane SRST and TRST initialization
At least some FT2232 based adapters don't necessarily come up
in the expected state, with SRST and TRST disabled. Since
other adapters could suffer the same problem, let's avoid
needing to patch every driver and just force *all* adapters
to initialize those values properly at server startup.
dbrownell [Wed, 7 Oct 2009 18:51:11 +0000 (18:51 +0000)]
Better fix for TAPs violating the JTAG spec for IR-Capture.
Instead of just assuming all IDCODE-deprived TAPs violate the
JTAG spec (they don't!), just require TAPs with such problems
to be declared with proper ircapture/irmask values. Example,
with mask and value of zero.
dbrownell [Wed, 7 Oct 2009 16:15:21 +0000 (16:15 +0000)]
Remove pointless "target library" chapter.
It had a very little bit of content; move that to the more extensive
chapter on config file guidelines, and give more current "ls" output
to show the available library code.
dbrownell [Wed, 7 Oct 2009 15:31:33 +0000 (15:31 +0000)]
Updates for "reset_config":
- revert to previous default: don't talk JTAG during SRST
- add "srst_nogates" flag, the converse of "srst_gates_jtag"
- with no args, display the current configuration
And update the User's Guide text with bullet lists to be a bit more clear.
dbrownell [Wed, 7 Oct 2009 05:43:05 +0000 (05:43 +0000)]
Change version labels to work better with GIT
- The guess-rev.sh script is now a tweaked version of "setlocalversion" as
seen in Linux, U-Boot, and various other projects. When it finds source
control support (git, hg, svn) it uses IDs from there. Else (specific
to this project) it reports itself as "-snapshot", e.g. from gitweb.
I verified this new "guess-rev.sh" script runs under Cygwin.
- Also update the generic version strings to be like "0.3.0-dev" (during
development) instead of the very long "0.3.0-in-development". These also
show up in the PDF docs. For better tracking, we might eventually change
these strings to include the version IDs too.
- Change the startup banner version strings so they include the guess-rev
output. Development and release versions with GIT will be like
Open On-Chip Debugger 0.3.0-dev-00282-g7191a4f-dirty (2009-10-05-20:57)
Open On-Chip Debugger 0.3.0 (2009-10-05-20:57)
instead of the previous SVN-specific (even when using git-svn!)
Open On-Chip Debugger 0.3.0-in-development (2009-10-05-01:39) svn:exported
Open On-Chip Debugger 0.3.0 (2009-10-05-01:39) Release
dbrownell [Mon, 5 Oct 2009 08:20:28 +0000 (08:20 +0000)]
Add a new JTAG "setup" event; use for better DaVinci ICEpick support.
The model is that this fires after scanchain verification, when it's
safe to call "jtag tapenable $TAPNAME". So it will fire as part of
non-error paths of "init" and "reset" command processing. However it
will *NOT* trigger during "jtag_reset" processing, which skips all
scan chain verification, or after verification errors.
ALSO:
- switch DaVinci chips to use this new mechanism
- log TAP activation/deactivation, since their IDCODEs aren't verified
- unify "enum jtag_event" scripted event notifications
- remove duplicative JTAG_TAP_EVENT_POST_RESET
dbrownell [Mon, 5 Oct 2009 08:18:17 +0000 (08:18 +0000)]
Minor cleanup to ARM926 debug entry:
- don't needlessly export this function
- handle "case 0" debug method-of-entry better (silent by default)
The "case 0" is a valid debug entry mode so it doesn't deserve the
warning int now gets. But it probably means that OpenOCD confused
itself somehow; or that it confused the ARM9EJS target.
dbrownell [Fri, 2 Oct 2009 09:19:03 +0000 (09:19 +0000)]
Minor ETB and ETM bugfixes and doc updates
- ETB
* report _actual_ hardware status, not just expected status
* add a missing diagnostic on a potential ETB setup error
* prefix any diagnostics with "ETB"
- ETM
* make "etm status" show ETM hardware status too, instead of
just traceport status (which previously was fake, sigh)
- Docs
* flesh out "etm tracemode" docs a bit
* clarify "etm status" ... previously it was traceport status
* explain "etm trigger_percent" as a *traceport* option
ETM+ETB tracing still isn't behaving, but now I can see that part of
the reason is that the ETB turns itself off almost immediately after
being enabled, and before collecting any data.
Update release process documentation.
- Improve and clarify the wording of the introduction.
- Add section on version taggging.
- Some other minor corrections.
Add numeric version tag support to release script.
The release process itself does not use this support yet, but it allows
packagers to automate the process of managing their own tags, if they
patch the source code before releasing binaries. The release processes
should be revised to incorporate this feature to support -rc packages.
Fix release script bugs after experience from 0.2.0:
- The NEWS step failed due to an empty commit comment.
- The final release step would have failed, because the steps to switch
from the secure to insecure repository (and back again) require both
switch and a URL relocation steps
- Don't issue needless JTAG resets ... only do them after
errors. Normal exit now leaves every TAP in BYPASS.
- Fix an unlikely memory leak on one fault path.
- Remove the oddball limitation that invalid capture LSBs
trigger errors only for TAPs that support IDCODE.
Re the JTAG reset: there are too many of them, and they can
(and do!) change system state. So the needless ones should
get removed. This one was especially pointless.
- Commands were supposed to have been "arm11 memwrite ..."
not "memwrite ..."
- Get rid of obfuscatory macros
- Re-alphabetize
- Add docs for "arm11 vcr"
Update FT2232 driver so that it reliably enters TAP_RESET.
When the OpenOCD server starts up it records its state as TAP_RESET,
even though it could be anything. Then when it starts to examine
the scan chain, it calls jtag_add_tlr() which sees it doesn't have
any work to do, and so it does nothing. This can make the next
operations fail because they start from the wrong TAP state...
Instead of caring about the current recorded state, always enter
TAP_RESET by forcing five clocks with TMS high.
(NOTE: it seems most other JTAG adapter drivers have this same bug.)
Streamline Capture-IR handling and integrity test.
Change the handling of the "-ircapture" and "-irmask" parameters
to be slightly more sensible, given that the JTAG spec describes
what is required, and that we already require that conformance in
one place. IR scan returns some bitstring with LSBs "01".
- First, provide and use default values that satisfy the IEEE spec.
Existing TAP configs will override the defaults, but those parms
are no longer required.
- Second, warn if any TAP gets set up to violate the JTAG spec.
It's likely a bug, but maybe not; else this should be an error.
Improve the related diagnostics to say which TAP is affected.
And associated minor fixes/cleanups to comments and diagnostics.
When attaching GDB to OpenOCD, the target state is no longer affected.
Added gdb_sync feature that allows GDB to sync up to target state.
Issue "monitor gdb_sync" and the next stepi, will return immediately
with updated register values to GDB.
Start handling the (second) SRST stage of reset better:
make sure that when there are two or more targets, their
various pre/post event reports are correctly ordered.
Previously, only the first target always saw its "pre"
method before SRST was asserted or deasserted.
When setting up an ETM, cache its ETM_CONFIG register. Then
only expose the registers which are actually present. They
could be missing for two basic reasons:
- This version might not support them at all; e.g. ETMv1.1
doesn't have some control/status registers. (My sample of
ARM9 boards shows all with ETMv1.3 support, FWIW.)
- The configuration on this chip may not populate as many
registers as possible; e.g. only two data value comparators
instead of eight.
Includes a bugfix in the "etm info" command: only one of the
two registers is missing on older silicon, so show the first
one before bailing.
Update ETM usage docs to explain that those registers need to be
written to configure what is traced, and that some ETM configs
are not yet handled. Also, give some examples of the kinds of
constrained trace which could be arranged.
Start cleaning up ETM register handling. On one ARM926 ETM+ETB
system, removes 20 non-existent registers ... but still includes
over 45 (!) ETM registers which don't even exist there ...
- Integrate the various tables to get one struct per register
- Get rid of needless per-register dynamic allocation
- Double check list of registers:
* Remove sixteen (!) non-registers for data comparators
* Remove four registers that imply newer ETM than we support
* Change some names to match current architecture specs
- Handle more register info
* some are write-only
* some are read-only
* record which versions have them, just in case
- Reorganize the registers to facilitate removing the extras
* group e.g. comparator/counter #N registers together
* add and use lookup-by-ID
- Add a header comment
- Line up the ETM context struct, pack it a bit
- Remove unused context_id (this doesn't support ETMv2 yet)
- Make most functions static
- Remove unused string table and other needless lines of code
- Correct "tracemode" helptext
Also provide and use an etm_reg_lookup() to find entries in the ETM
register cache. This will help cope with corrected contents of that
cache, which doesn't include entires for non-existent registers.
projects / openocd / log