The restored boot registration block in branch sched/508-boot-thread-disk-wait at b346196295db73f9fe4b44c8f080e75217f27693 reaches argv_test, which fails because its initial argument vector is empty.
Evidence from the x86 boot_tests,testing,external_test_bins profile:
serial_user.txt from the repaired boot reports argc: 0, FAIL: argc is 0 (expected at least 1), and ARGV_TEST_FAILED.serial_kernel.txt records process 155, named argv_test, exiting with code 1.- The same boot's
serial_user.txt reports [TEST_EXEC_BLOCK_COMPLETE:x86:calls=64] and [BOOT_DISK_WAIT_ORACLE:x86:switched_away=0:tests_completed=1:PASS].
At the cited source revision, kernel/src/process/creation.rs:42 calls manager.create_process(name.clone(), elf_data) on x86. Its aarch64 counterpart in the same file prepares a NUL-terminated default argv[0] and calls create_process_with_argv. kernel/src/test_exec.rs:3223 uses the x86 wrapper to launch this test, and userspace/programs/src/argv_test.rs requires at least one argument.
The full x86 gate also stops before a terminal tally with an ext2-lock stall; that observation is separate from this reproducible argv failure. This issue tracks setting up a valid default argv[0] for direct x86 process creation and retaining the existing argv test's checks. Do not make the test accept argc=0.
The 508 round's captures are under docs/planning/green-program/block/serials/508/repaired-boot/ on that branch. The source change is pushed; the evidence is being retained with the round's documentation commit.
The restored boot registration block in branch
sched/508-boot-thread-disk-waitatb346196295db73f9fe4b44c8f080e75217f27693reachesargv_test, which fails because its initial argument vector is empty.Evidence from the x86
boot_tests,testing,external_test_binsprofile:serial_user.txtfrom the repaired boot reportsargc: 0,FAIL: argc is 0 (expected at least 1), andARGV_TEST_FAILED.serial_kernel.txtrecords process 155, namedargv_test, exiting with code 1.serial_user.txtreports[TEST_EXEC_BLOCK_COMPLETE:x86:calls=64]and[BOOT_DISK_WAIT_ORACLE:x86:switched_away=0:tests_completed=1:PASS].At the cited source revision,
kernel/src/process/creation.rs:42callsmanager.create_process(name.clone(), elf_data)on x86. Its aarch64 counterpart in the same file prepares a NUL-terminated default argv[0] and callscreate_process_with_argv.kernel/src/test_exec.rs:3223uses the x86 wrapper to launch this test, anduserspace/programs/src/argv_test.rsrequires at least one argument.The full x86 gate also stops before a terminal tally with an ext2-lock stall; that observation is separate from this reproducible argv failure. This issue tracks setting up a valid default argv[0] for direct x86 process creation and retaining the existing argv test's checks. Do not make the test accept argc=0.
The 508 round's captures are under
docs/planning/green-program/block/serials/508/repaired-boot/on that branch. The source change is pushed; the evidence is being retained with the round's documentation commit.