| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
|
|
| |
fake.dts was changed to have fixed addresses for core (to be able to
test exact dn name match).
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Tested-by: Rashmica Gupta <rashmica.g@gmail.com>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Tested-by: Rashmica Gupta <rashmica.g@gmail.com>
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
|
|
|
|
|
| |
Restores the functionality that allows to set I2C backend
device through command line arguments, it was broken by commit
dacc9b57e4615f125eb1f56d842a64c40abb1408
Signed-off-by: Artem Senichev <artemsen@gmail.com>
Reviewed-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Not all memory can be read with the default ADU block size of 8
bytes. Specifically cache-inhibited access to some MMIO regions such
as PCIe BAR spaces requires 4 byte accesses to avoid check stopping
the machine.
This patch adds library functions to read/write IO memory which takes
a block size argument and two new put/getmemio commands to allow a
specific block size to be specified from the command line.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Reviewed-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
| |
Signed-off-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Rashmica Gupta <rashmica.g@gmail.com>
|
|
|
|
|
|
|
|
|
| |
If none of the expected targets are selected, log an error.
core htm - core target
nest htm - pib target
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
|
|
|
| |
Parsing -l options requires device_node to be set. It is used in
pir_map() to map Linux CPU to proc/core/thread.
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
|
|
|
|
|
| |
It's possible that pdbg doesn't have hw units defined for all the nodes
in the device tree. So if we encounter a node without class name ignore
that node while matching path.
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
|
| |
With capability of selecting any target in the device tree, probe output
should display all the nodes with their device tree names.
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
|
|
| |
When pdbg is built without "--disable-shared" option, libtool creates
pdbg as a wrapper around lt-pdbg binary. This causes the tests to fail
which try to match the output containing argv[0].
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
| |
Drop the old target selection code.
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
|
| |
To maintain the compatibility with the old selection where -p selected
processor, add fsi and pib targets with matching index.
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
| |
... and document the function.
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
| |
Signed-off-by: Amitay Isaacs <amitay@ozlabs.org>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 936dbdcedb27 ("libpdbg: Rework target addressing") introduced a
bug leading to the following assertion failing:
pdbg: libpdbg/device.c:634: pdbg_target_address: Assertion `(pos + n) <= p->len' failed.
When this function was reworked the index parameter was dropped as
every caller set index == 0. Removal should have also resulted in the
local pos variable being removed. Instead it was set as if index == 1
resulting in the above violation.
Fix the bug and add a test to check pdbg_target_address().
Signed-off-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
| |
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Reviewed-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
|
| |
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Reviewed-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
|
|
|
|
|
|
| |
Some of the functions dealing with target properties have somewhat
inconsistent naming. This patch renames them and adds #defines for
backwards compatibility for external projects. These will be removed
once older projects have moved over to the new names.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Reviewed-by: Amitay Isaacs <amitay@ozlabs.org>
|
|
|
|
|
| |
Signed-off-by: Rashmica Gupta <rashmica.g@gmail.com>
Signed-off-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
|
| |
Add in the generated c file so if the build system doesn't have ragel gdbserver can still
be built.
Signed-off-by: Rashmica Gupta <rashmica.g@gmail.com>
Signed-off-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
|
| |
I have changed a few bits here and there but this patch is largely
authored by Alistair Popple.
Signed-off-by: Rashmica Gupta <rashmica.g@gmail.com>
Signed-off-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The stack unwinder currently does not do any endian conversion, which
means it won't work correctly if the stack does not match pdbg endian.
This patch attempts an endian flip if the stack looks wrong, and goes
with that if it's an improvement. It also has some magic hackery to
take OPAL->Linux into account. Unfortunately this is not a "clean"
fully general solution, but works reasonably well in practice.
This is the regs --backtrace output for a test that has a CPU hang in
an OPAL call from Linux:
STACK: SP NIA
0x0000000031c43cb0 0x000000003002b324 (big-endian)
0x0000000031c43d20 0x00000000300051e4 (big-endian)
0xc000200006283b60 0xc00000000008f1c8 (little-endian)
0xc000200006283c40 0xc00000000002af18 (little-endian)
0xc000200006283c70 0xc000000000114064 (little-endian)
0xc000200006283ce0 0xc0000000001144d0 (little-endian)
0xc000200006283e30 0xc00000000000b288 (little-endian)
0x00007fffe28d0cb0
We can see the stack unwind from OPAL to Linux to userspace (which
does not get decoded -- yet).
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
CI is failing to build:
src/options_arm.c:65:3: error: ignoring return value of 'fscanf',
declared with attribute warn_unused_result [-Werror=unused-result]
fscanf(cfam_id_file, "0x%" PRIx32, &cfam_id);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
If we don't get the expected value in fscanf, display an error. The
code already will do the correct thing and fall through to the unknown
device case.
Signed-off-by: Joel Stanley <joel@jms.id.au>
|
|
|
|
|
|
|
|
|
|
|
| |
The backtrace causes us to read memory which appears to have a high
likelihood of causing a checkstop due to an invalid address.
This adds a --backtrace flag to the regs command which enables the
backtrace. By default it is off to protect the user.
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
|
| |
Defines a new macro allowing commands that only taking flags and no positional
arguments to be defined. Also adds a couple of test cases for the new macro.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Joel Stanley <joel@jms.id.au>
|
|
|
|
| |
Signed-off-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If a user explicitly specifies a backend assume they know what they're doing and
use it. If no backend is specified pick some sane defaults based on the detected
platform.
Fixes the following bug when specifying a backend is incorrectly determined to
be invalid:
Target p8 not possible
kernel: No target is necessary
i2c: No target is necessary
fsi: p8 p9w p9r p9z
Signed-off-by: Alistair Popple <alistair@popple.id.au>
|
|
|
|
| |
Signed-off-by: Rashmica Gupta <rashmica.g@gmail.com>
|
|
|
|
|
| |
Signed-off-by: Rashmica Gupta <rashmica.g@gmail.com>
Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
|
|
|
|
|
| |
Signed-off-by: Rashmica Gupta <rashmica.g@gmail.com>
Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
|
|
|
|
|
| |
Signed-off-by: Rashmica Gupta <rashmica.g@gmail.com>
Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
|