On 6/28/22 9:16 PM, Adam Sindelar wrote: > On Tue, Jun 28, 2022 at 08:16:41PM +0530, Aneesh Kumar K V wrote: >> On 6/28/22 12:22 AM, Andrew Morton wrote: >>> The patch titled >>> Subject: selftests/vm: Only run 128TBswitch with 5-level paging >>> has been added to the -mm mm-unstable branch. Its filename is >>> selftests-vm-only-run-128tbswitch-with-5-level-paging.patch >>> >>> This patch will shortly appear at >>> https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/selftests-vm-only-run-128tbswitch-with-5-level-paging.patch >>> >>> This patch will later appear in the mm-unstable branch at >>> git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm >>> >>> Before you just go and hit "reply", please: >>> a) Consider who else should be cc'ed >>> b) Prefer to cc a suitable mailing list as well >>> c) Ideally: find the original patch on the mailing list and do a >>> reply-to-all to that, adding suitable additional cc's >>> >>> *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** >>> >>> The -mm tree is included into linux-next via the mm-everything >>> branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm >>> and is updated there every 2-3 working days >>> >>> ------------------------------------------------------ >>> From: Adam Sindelar <adam@xxxxxxxxxxxx> >>> Subject: selftests/vm: Only run 128TBswitch with 5-level paging >>> Date: Mon, 27 Jun 2022 18:39:12 +0200 >>> >>> The test va_128TBswitch.c expects to be able to pass mmap an address hint >>> and length that cross the address 1<<47. This is not possible without >>> 5-level page tables, so the test fails. >>> >>> The test is already only run on 64-bit powerpc and x86 archs, but this >>> patch adds an additional check that skips the test if PG_TABLE_LEVELS < 5. >>> There is precedent for checking /proc/config.gz in selftests, e.g. in >>> selftests/firmware. >>> >>> Running the tests produces the desired output: >>> >>> sudo make -C tools/testing/selftests TARGETS=vm run_tests >>> --------------------------- >>> running ./va_128TBswitch.sh >>> --------------------------- >>> ./va_128TBswitch.sh: PG_TABLE_LEVELS=4, must be >= 5 to run this test >>> [SKIP] >>> ------------------------------- >>> >>> Link: https://lkml.kernel.org/r/20220627163912.5581-1-adam@xxxxxxxxxxxx >>> Signed-off-by: Adam Sindelar <adam@xxxxxxxxxxxx> >>> Cc: Adam Sindelar <ats@xxxxxx> >>> Cc: David Vernet <void@xxxxxxxxxxxxx> >>> Cc: Aneesh Kumar K.V <aneesh.kumar@xxxxxxxxxxxxxxxxxx> >>> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> >>> --- >>> >>> tools/testing/selftests/vm/Makefile | 1 >>> tools/testing/selftests/vm/run_vmtests.sh | 2 >>> tools/testing/selftests/vm/va_128TBswitch.sh | 39 +++++++++++++++++ >>> 3 files changed, 41 insertions(+), 1 deletion(-) >>> >>> --- a/tools/testing/selftests/vm/Makefile~selftests-vm-only-run-128tbswitch-with-5-level-paging >>> +++ a/tools/testing/selftests/vm/Makefile >>> @@ -93,6 +93,7 @@ TEST_PROGS := run_vmtests.sh >>> >>> TEST_FILES := test_vmalloc.sh >>> TEST_FILES += test_hmm.sh >>> +TEST_FILES += va_128TBswitch.sh >>> >>> KSFT_KHDR_INSTALL := 1 >>> include ../lib.mk >>> --- a/tools/testing/selftests/vm/run_vmtests.sh~selftests-vm-only-run-128tbswitch-with-5-level-paging >>> +++ a/tools/testing/selftests/vm/run_vmtests.sh >>> @@ -158,7 +158,7 @@ if [ $VADDR64 -ne 0 ]; then >>> run_test ./virtual_address_range >>> >>> # virtual address 128TB switch test >>> - run_test ./va_128TBswitch >>> + run_test ./va_128TBswitch.sh >>> fi # VADDR64 >>> >>> # vmalloc stability smoke test >>> --- /dev/null >>> +++ a/tools/testing/selftests/vm/va_128TBswitch.sh >>> @@ -0,0 +1,39 @@ >>> +#!/bin/bash >>> +# SPDX-License-Identifier: GPL-2.0 >>> +# >>> +# Copyright (C) 2022 Adam Sindelar (Meta) <adam@xxxxxxxxxxxx> >>> +# >>> +# This is a test for mmap behavior with 5-level paging. This script wraps the >>> +# real test to check that the kernel is configured to support at least 5 >>> +# pagetable levels. >>> + >>> +# 1 means the test failed >>> +exitcode=1 >>> + >>> +# Kselftest framework requirement - SKIP code is 4. >>> +ksft_skip=4 >>> + >>> +die() >>> +{ >>> + echo "$1" >>> + exit $exitcode >>> +} >>> + >>> +check_test_requirements() >>> +{ >>> + local config="/proc/config.gz" >>> + [[ -f "${config}" ]] || config="/boot/config-$(uname -r)" >>> + [[ -f "${config}" ]] || die "Cannot find kernel config in /proc or /boot" >>> + >>> + # gzip -dcfq automatically handles both compressed and plaintext input. >>> + # See man 1 gzip under '-f'. >>> + local pg_table_levels=$(gzip -dcfq "${config}" | grep PGTABLE_LEVELS | cut -d'=' -f 2) >>> + >>> + if [[ "${pg_table_levels}" -lt 5 ]]; then >>> + echo "$0: PG_TABLE_LEVELS=${pg_table_levels}, must be >= 5 to run this test" >>> + exit $ksft_skip >>> + fi >>> +} >>> + >> >> What about architectures other than x86? Will this break on other architectures which don't support >> PGTABLE_LEVELS ? >> >> > > The test only runs on powerpc64 and x86_64. I didn't realize powerpc64 > is using fewer pagetable levels. I am inclined to just change v4 so that > it checks the PGTABLE_LEVELS config on x86_64 and accepts any powerpc64, > because I don't know how to check whether a specific powerpc64 system > supports addresses over 1<<47. (The test already rejects all other > architectures.) > > Does that seem reasonable? I would suggest restricting the new check to only x86 -aneesh
