Traps tlb test kernel page new#32
Open
stzahi1 wants to merge 3 commits into
Open
Conversation
Signed-off-by: Tzahi Sabo <stzahi@qti.qualcomm.com>
scripts/Makefile.coverage extracts the per-ARCHV test list with the
perl filter s/^\s*\#.*//; which only removes comments that start at
the beginning of a line. testlist.v81 line 59 has a trailing comment:
./kernel/traps/tlb/test #FIXME: needs to adapt to ...
so SUBDIRS ended up containing the literal tokens "#FIXME:", "needs",
"to", etc. The colon in "#FIXME:" then leaked through the patsubst
on line 17 into a prerequisite of "tst:", which make parses as a
static-pattern rule and rejects with:
scripts/Makefile.coverage:27: *** multiple target patterns. Stop.
This broke every target in the file, including \`make clean\` /
\`make covclean\`, because the error fires at parse time.
Drop the ^ anchor so the regex strips inline comments as well as
full-line ones.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Signed-off-by: Tzahi Sabo <stzahi@qti.qualcomm.com>
The test had two latent bugs that the recent V81 enable (commit 5d13ae7e) exposed. 1. Hardcoded JTLB layout. The test passed last_tlb_index=125 and tlb_size=128 to H2K_kg_init, and used 64/63/65 as the allocator floor. These values only match a 128-entry JTLB. V81 reports tlb_size=192 via CFG_TABLE_JTLB_SIZE, so the synthesized kg state diverged from the actual hardware and the allocator's count assertion failed. 2. (u32_t)&H2K_KERNEL_NPAGES is meaningless for STANDALONE tests. stake/consts.pl computes kernel_npages assuming the ELF is linked at H2K_LINK_ADDR (the real kernel layout). For a STANDALONE test ELF that lives elsewhere, the formula wraps around -- on V81 it produced 0xffffffffffffc043, blowing up any derivation that used it. Fix: - Read tlb_size at runtime from H2K_cfg_table(CFG_TABLE_JTLB_SIZE). - Pick a fixed TEST_PINNED_COUNT=2 (synthetic test fixture; the test has no kernel image to map, so the production npages+device+angel formula doesn't apply). last_tlb_index = tlb_size - 2 - 1. - Replace every "64" allocator-floor literal with (tlb_size - ALLOC_WINDOW), where ALLOC_WINDOW=64 is the architectural width of pinned_tlb_mask. Off-by-one neighbors become ALLOC_WINDOW +/- 1. - Generalize test_readwriteprobe's index range from the hardcoded 32..96 to (tlb_size - 3*ALLOC_WINDOW/2)..(tlb_size - ALLOC_WINDOW/2). This is byte-identical to 32..96 for tlb_size=128 (preserving legacy v60/v5/v4 behavior) and now actually crosses the allocator boundary on V81. - Replace the mask-bit-width literal 0x3F with (ALLOC_WINDOW - 1). - Replace the page-shift literal 12 in VALID_VA with PAGE_BITS. - Name (32+20) as TLB_ASID_SHIFT 52 with a pointer to tlbfmt.h. - Promote the pinned-slot marker base 0x01234567cafe0000ULL to PINNED_MARKER_BASE. - Add header comments to TLBCONST, VALID_VPN, ASID, VALID_TLB_ENTRY, TEST_PINNED_COUNT, ALLOC_WINDOW explaining what each value represents and why it was chosen. - Drop the now-obsolete FIXME from scripts/testlist.v81 -- the test now adapts to any tlb_size, so the "needs to adapt to different kernel page size/count" caveat no longer applies. Include changes: - Add <cfg_table.h> for H2K_cfg_table and CFG_TABLE_JTLB_SIZE. Skip <symbols.h>: it declares H2K_KERNEL_NPAGES (no longer needed) and H2K_ALLOC_HEAP_SIZE, the latter of which collides with the test build's -DH2K_ALLOC_HEAP_SIZE CFLAG. Verified: v81/opt and v81/ref both pass 92 tests. The math reduces to the original 32..96 range for tlb_size=128, so the v60/v5/v4 testlist entries are unaffected. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> Signed-off-by: Tzahi Sabo <stzahi@qti.qualcomm.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.