baseport/remoteproc: Make ADSP/CDSP/GPDSP/WPSS tests SoC-aware, add RPMsg probe, harden logs & shellcheck #130
Conversation
| acquire_test_lock "$TESTNAME" | ||
| log_info "Starting $TESTNAME Testcase" | ||
|
|
||
| #Discover |
There was a problem hiding this comment.
I don't think we should define the test plan / test cases to run by dynamically discover what's available on the platform. We are hiding a class of issues by doing that, especially all issues that lead to the device not being functional on a booted system.
There was a problem hiding this comment.
I don't think we should define the test plan / test cases to run by dynamically discover what's available on the platform. We are hiding a class of issues by doing that, especially all issues that lead to the device not being functional on a booted system.
I’ve pushed the latest updates to read the DT-populated entries and then check for probing, as you recommended.
smuppand
changed the title
|
|
||
| [ -n "$fw" ] || return 3 | ||
|
|
||
| found=0 |
There was a problem hiding this comment.
AFAIK found is not required anywhere
There was a problem hiding this comment.
AFAIK found is not required anywhere
the found variable was used as a simple flag to record whether any of the directory‐based “fast-path” checks succeeded
found=0 for d in ...; do [ -d "$d" ] && found=1 && break done [ "$found" -eq 1 ] && return 0
-
Initializes found=0.
-
Loops over potential DT directories (e.g. /proc/device-tree/smp2p-*).
-
If any directory exists, sets found=1 and breaks out of the loop.
-
After the loop, checks found to decide whether to return success.
smuppand
force-pushed
the
rproc
branch
2 times, most recently
from
195a598 to
d268dc4
Compare
…Msg probe, shellcheck fixes - Discover RProc nodes via find_remoteproc_by_firmware() instead of hardcoded IDs. - Skip cleanly when a core isn’t present on the SoC (no more false FAILs). - Stop/Start paths now wait_for_state()/wait_remoteproc_state() with timeouts & polling. - Dump rproc logs around each transition (before/after stop/start) for easier CI debug. - Add generic RPMsg exercise: * Map rpmsg_ctrl* to the owning remoteproc. * Reuse/create endpoints and try an echo; mark PASS/FAIL/SKIPPED accordingly. - CDSP: handle multi-instance (cdsp0/cdsp1…/gpdsp0/gpdsp1...) and print per-instance boot/stop/start/ping results. - WPSS: keep driver fallback (ath11k/ath11xx) if remoteproc is absent: * Validate firmware via /lib/firmware/ath11k blobs, dmesg strings, and wlan*/ath* netdev presence. * POSIX-safe globbing instead of grep-regex for interface detection. - Ensure .res contains only “TESTNAME <RESULT>”; everything else goes to log. - ShellCheck cleanup: SC2143, SC2181, SC3037, SC2034, etc. (no unused vars, no `$?` chains, printf over echo -e). - Added/updated helpers in functestlib.sh (rpmsg_* helpers, dump_rproc_logs, wait_remoteproc_state, etc.). - Protect concurrent runs with acquire_test_lock()/release_test_lock(). - Added DT check for each adsp/cdsp/wpss tests - Added dt_has_remoteproc_fw to read the /proc/device-tree for the given fw. Signed-off-by: Srikanth Muppandam <smuppand@qti.qualcomm.com>
3 participants
This PR refactors and toughens the remoteproc test scripts for ADSP, CDSP (multi-instance), and WPSS while keeping backward compatibility with existing functestlib helpers.
SoC-aware discovery & skips
Robust stop/start sequencing
Generic RPMsg exercise (optional)
CDSP/GPDSP multi-instance reporting
WPSS driver fallback kept & improved
Concurrency safety