- Linux-kselftest-mirror - lists.linaro.org

selftests: sigaltstack: sas # exit=1 - # Bail out! SP is not on sigaltstack - on clang build

by Naresh Kamboju

kselftest: sigaltstack built with clang-16 getting failed but passed with gcc-12 build. Please find more details about test logs on clang-16 and gcc-12 and steps to reproduce locally on your machine by using tuxrun. Reported-by: Linux Kernel Functional Testing <lkft(a)linaro.org> Test log: ---------- Linux version 6.3.0-rc1-next-20230307 (tuxmake@tuxmake) (Debian clang version 16.0.0 (++20230228093516+60692a66ced6-1~exp1~20230228093525.41), Debian LLD 16.0.0) #1 SMP PREEMPT @1678159722 ... kselftest: Running tests in sigaltstack TAP version 13 1..1 # selftests: sigaltstack: sas # # [NOTE] the stack size is 21104 # TAP version 13 # 1..3 # ok 1 Initial sigaltstack state was SS_DISABLE # Bail out! SP is not on sigaltstack # # Planned tests != run tests (3 != 1) # # Totals: pass:1 fail:0 xfail:0 xpass:0 skip:0 error:0 not ok 1 selftests: sigaltstack: sas # exit=1 Links, qemu-x86_64: - https://qa-reports.linaro.org/lkft/linux-next-master/build/next-20230307/te… qemu-arm64: - https://qa-reports.linaro.org/lkft/linux-next-master/build/next-20230307/te… - https://qa-reports.linaro.org/lkft/linux-next-master/build/next-20230307/te… kselftest: sigaltstack built with gcc-12 getting pass. Test log: --------- Linux version 6.3.0-rc1-next-20230307 (tuxmake@tuxmake) (aarch64-linux-gnu-gcc (Debian 12.2.0-14) 12.2.0, GNU ld (GNU Binutils for Debian) 2.40) #1 SMP PREEMPT @1678159736 ... kselftest: Running tests in sigaltstack TAP version 13 1..1 # selftests: sigaltstack: sas # # [NOTE] the stack size is 50080 # TAP version 13 # 1..3 # ok 1 Initial sigaltstack state was SS_DISABLE # # [RUN] signal USR1 # ok 2 sigaltstack is disabled in sighandler # # [RUN] switched to user ctx # # [RUN] signal USR2 # # [OK] Stack preserved # ok 3 sigaltstack is still SS_AUTODISARM after signal # # Totals: pass:3 fail:0 xfail:0 xpass:0 skip:0 error:0 ok 1 selftests: sigaltstack: sas Links, qemu-x86_64: - https://qa-reports.linaro.org/lkft/linux-next-master/build/next-20230307/te… qemu-arm64: - https://qa-reports.linaro.org/lkft/linux-next-master/build/next-20230307/te… - https://qa-reports.linaro.org/lkft/linux-next-master/build/next-20230307/te… Steps to reproduce: -------------- # To install tuxrun on your system globally: # sudo pip3 install -U tuxrun==0.37.2 # # See https://tuxrun.org/ for complete documentation. tuxrun \ --runtime podman \ --device qemu-x86_64 \ --boot-args rw \ --kernel https://storage.tuxsuite.com/public/linaro/lkft/builds/2MfXLbLsvBufVn1MrsnX… \ --modules https://storage.tuxsuite.com/public/linaro/lkft/builds/2MfXLbLsvBufVn1MrsnX… \ --rootfs https://storage.tuxboot.com/debian/bookworm/amd64/rootfs.ext4.xz \ --parameters SKIPFILE=skipfile-lkft.yaml \ --parameters KSELFTEST=https://storage.tuxsuite.com/public/linaro/lkft/builds/2MfXLbLsvB… \ --image docker.io/lavasoftware/lava-dispatcher:2023.01.0020.gc1598238f \ --tests kselftest-sigaltstack \ --timeouts boot=15 -- Linaro LKFT https://lkft.linaro.org

2 years, 3 months

4
4
0 0

[PATCH v3 0/5] mm: userfaultfd: refactor and add UFFDIO_CONTINUE_MODE_WP

by Axel Rasmussen

This series, currently based on 6.3-rc1, is divided into two parts: - Commits 1-4 refactor userfaultfd ioctl code without behavior changes, with the main goal of improving consistency and reducing the number of function args. - Commit 5 adds UFFDIO_CONTINUE_MODE_WP. The refactors are sorted by increasing controversial-ness, the idea being we could drop some of the refactors if they are deemed not worth it. Changelog: v2->v3: - rebase onto 6.3-rc1 - typedef a new type for mfill flags in patch 3/5 (suggested by Nadav) v1->v2: - refactor before adding the new flag, to avoid perpetuating messiness Axel Rasmussen (5): mm: userfaultfd: rename functions for clarity + consistency mm: userfaultfd: don't pass around both mm and vma mm: userfaultfd: combine 'mode' and 'wp_copy' arguments mm: userfaultfd: don't separate addr + len arguments mm: userfaultfd: add UFFDIO_CONTINUE_MODE_WP to install WP PTEs fs/userfaultfd.c | 120 +++++------- include/linux/hugetlb.h | 27 ++- include/linux/shmem_fs.h | 9 +- include/linux/userfaultfd_k.h | 61 +++--- include/uapi/linux/userfaultfd.h | 7 + mm/hugetlb.c | 34 ++-- mm/shmem.c | 14 +- mm/userfaultfd.c | 235 +++++++++++------------ tools/testing/selftests/mm/userfaultfd.c | 4 + 9 files changed, 247 insertions(+), 264 deletions(-) -- 2.40.0.rc0.216.gc4246ad0f0-goog

2 years, 3 months

4
18
0 0

[PATCH 2/2] selftests/clone3: test clone3 with CLONE_NEWTIME

by Tobias Klauser

Verify that clone3 can be called successfully with CLONE_NEWTIME in flags. Cc: Andrey Vagin <avagin(a)openvz.org> Cc: Christian Brauner <brauner(a)kernel.org> Cc: Shuah Khan <shuah(a)kernel.org> Signed-off-by: Tobias Klauser <tklauser(a)distanz.ch> --- tools/testing/selftests/clone3/clone3.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/tools/testing/selftests/clone3/clone3.c b/tools/testing/selftests/clone3/clone3.c index cd4582129c7d..4fce46afe6db 100644 --- a/tools/testing/selftests/clone3/clone3.c +++ b/tools/testing/selftests/clone3/clone3.c @@ -195,5 +195,8 @@ int main(int argc, char *argv[]) test_clone3(CLONE_NEWPID, getpagesize() + 8, -E2BIG, CLONE3_ARGS_NO_TEST); + /* Do a clone3() in a new time namespace */ + test_clone3(CLONE_NEWTIME, 0, 0, CLONE3_ARGS_NO_TEST); + return !ksft_get_fail_cnt() ? ksft_exit_pass() : ksft_exit_fail(); } -- 2.39.1

2 years, 3 months

1
1
0 0

[PATCH v1 0/2] KVM: selftests: Fixes for broken tests

by Ryan Roberts

During the course of implementing FEAT_LPA2 within the arm64 KVM port, I found a couple of issues within the KVM selftest code, which I thought were worth posting independently. The LPA2 patches, for which I will post v2 in the next few days, depend on these fixes for its testing. Ryan Roberts (2): KVM: selftests: Fixup config fragment for access_tracking_perf_test KVM: selftests: arm64: Fix pte encode/decode for PA bits > 48 tools/testing/selftests/kvm/config | 1 + .../selftests/kvm/lib/aarch64/processor.c | 32 ++++++++++++++----- 2 files changed, 25 insertions(+), 8 deletions(-) -- 2.25.1

2 years, 3 months

2
5
0 0

[PATCH] selftests: pci: pci-selftest: add support for PCI endpoint driver test

by Aman Gupta

This patch enables the support to perform selftest on PCIe endpoint driver present in the system. The following tests are currently performed by the selftest utility 1. BAR Tests (BAR0 to BAR5) 2. MSI Interrupt Tests (MSI1 to MSI32) 3. Read Tests (For 1, 1024, 1025, 1024000, 1024001 Bytes) 4. Write Tests (For 1, 1024, 1025, 1024000, 1024001 Bytes) 5. Copy Tests (For 1, 1024, 1025, 1024000, 1024001 Bytes) Signed-off-by: Aman Gupta <aman1.gupta(a)samsung.com> Signed-off-by: Padmanabhan Rajanbabu <p.rajanbabu(a)samsung.com> --- tools/testing/selftests/Makefile | 1 + tools/testing/selftests/pci/.gitignore | 1 + tools/testing/selftests/pci/Makefile | 7 + tools/testing/selftests/pci/pci-selftest.c | 167 +++++++++++++++++++++ 4 files changed, 176 insertions(+) create mode 100644 tools/testing/selftests/pci/.gitignore create mode 100644 tools/testing/selftests/pci/Makefile create mode 100644 tools/testing/selftests/pci/pci-selftest.c diff --git a/tools/testing/selftests/Makefile b/tools/testing/selftests/Makefile index c2064a35688b..81584169a80f 100644 --- a/tools/testing/selftests/Makefile +++ b/tools/testing/selftests/Makefile @@ -49,6 +49,7 @@ TARGETS += net/forwarding TARGETS += net/mptcp TARGETS += netfilter TARGETS += nsfs +TARGETS += pci TARGETS += pidfd TARGETS += pid_namespace TARGETS += powerpc diff --git a/tools/testing/selftests/pci/.gitignore b/tools/testing/selftests/pci/.gitignore new file mode 100644 index 000000000000..db01411b8200 --- /dev/null +++ b/tools/testing/selftests/pci/.gitignore @@ -0,0 +1 @@ +pci-selftest diff --git a/tools/testing/selftests/pci/Makefile b/tools/testing/selftests/pci/Makefile new file mode 100644 index 000000000000..76b7725a45ae --- /dev/null +++ b/tools/testing/selftests/pci/Makefile @@ -0,0 +1,7 @@ +# SPDX-License-Identifier: GPL-2.0 +CFLAGS += -O2 -Wl,-no-as-needed -Wall +LDFLAGS += -lrt -lpthread -lm + +TEST_GEN_PROGS = pci-selftest + +include ../lib.mk diff --git a/tools/testing/selftests/pci/pci-selftest.c b/tools/testing/selftests/pci/pci-selftest.c new file mode 100644 index 000000000000..73e8f3eb1982 --- /dev/null +++ b/tools/testing/selftests/pci/pci-selftest.c @@ -0,0 +1,167 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * PCI Endpoint Driver Test Program + * + * Copyright (c) 2022 Samsung Electronics Co., Ltd. + * https://www.samsung.com + * Author: Aman Gupta <aman1.gupta(a)samsung.com> + */ + +#include <errno.h> +#include <fcntl.h> +#include <stdbool.h> +#include <stdio.h> +#include <stdlib.h> +#include <sys/ioctl.h> +#include <unistd.h> + +#include "../kselftest_harness.h" + +#define PCITEST_BAR _IO('P', 0x1) +#define PCITEST_LEGACY_IRQ _IO('P', 0x2) +#define PCITEST_MSI _IOW('P', 0x3, int) +#define PCITEST_WRITE _IOW('P', 0x4, unsigned long) +#define PCITEST_READ _IOW('P', 0x5, unsigned long) +#define PCITEST_COPY _IOW('P', 0x6, unsigned long) +#define PCITEST_MSIX _IOW('P', 0x7, int) +#define PCITEST_SET_IRQTYPE _IOW('P', 0x8, int) +#define PCITEST_GET_IRQTYPE _IO('P', 0x9) +#define PCITEST_CLEAR_IRQ _IO('P', 0x10) + +static char *test_device = "/dev/pci-endpoint-test.0"; + +struct xfer_param { + unsigned long size; + unsigned char flag; + }; + +FIXTURE(device) +{ + int fd; +}; + +FIXTURE_SETUP(device) +{ + + self->fd = open(test_device, O_RDWR); + + ASSERT_NE(-1, self->fd) { + TH_LOG("Can't open PCI Endpoint Test device\n"); + } +} + +FIXTURE_TEARDOWN(device) +{ + close(self->fd); +} + +TEST_F(device, BAR_TEST) +{ + int ret = -EINVAL; + int final = 0; + + for (int i = 0; i <= 5; i++) { + ret = ioctl(self->fd, PCITEST_BAR, i); + + EXPECT_EQ(1, ret) { + TH_LOG("TEST FAILED FOR BAR %d\n", i); + final++; + } + } + + ASSERT_EQ(0, final); +} + +TEST_F(device, MSI_TEST) +{ + int ret = -EINVAL; + int final = 0; + + ret = ioctl(self->fd, PCITEST_SET_IRQTYPE, 1); + ASSERT_EQ(1, ret); + + for (int i = 1; i <= 32; i++) { + ret = ioctl(self->fd, PCITEST_MSI, i); + EXPECT_EQ(1, ret) { + TH_LOG("TEST FAILED FOR MSI%d\n", i); + final++; + } + } + + ASSERT_EQ(0, final); +} + +TEST_F(device, READ_TEST) +{ + int final = 0; + int ret = -EINVAL; + unsigned long SIZE[5] = {1, 1024, 1025, 1024000, 1024001}; + + ret = ioctl(self->fd, PCITEST_SET_IRQTYPE, 1); + ASSERT_EQ(1, ret); + + struct xfer_param param; + + param.flag = 0; + for (int i = 0; i < 5; i++) { + param.size = SIZE[i]; + ret = ioctl(self->fd, PCITEST_READ, &param); + EXPECT_EQ(1, ret) { + TH_LOG("TEST FAILED FOR size =%ld.\n", SIZE[i]); + final++; + } + } + + ASSERT_EQ(0, final); +} + +TEST_F(device, WRITE_TEST) +{ + int final = 0; + int ret = -EINVAL; + unsigned long SIZE[5] = {1, 1024, 1025, 1024000, 1024001}; + + ret = ioctl(self->fd, PCITEST_SET_IRQTYPE, 1); + ASSERT_EQ(1, ret); + + struct xfer_param param; + + param.flag = 0; + + for (int i = 0; i < 5; i++) { + param.size = SIZE[i]; + ret = ioctl(self->fd, PCITEST_WRITE, &param); + EXPECT_EQ(1, ret) { + TH_LOG("TEST FAILED FOR size =%ld.\n", SIZE[i]); + final++; + } + } + + ASSERT_EQ(0, final); +} + +TEST_F(device, COPY_TEST) +{ + int final = 0; + int ret = -EINVAL; + unsigned long SIZE[5] = {1, 1024, 1025, 1024000, 1024001}; + + ret = ioctl(self->fd, PCITEST_SET_IRQTYPE, 1); + ASSERT_EQ(1, ret); + + struct xfer_param param; + + param.flag = 0; + + for (int i = 0; i < 5; i++) { + param.size = SIZE[i]; + ret = ioctl(self->fd, PCITEST_COPY, &param); + EXPECT_EQ(1, ret) { + TH_LOG("TEST FAILED FOR size =%ld.\n", SIZE[i]); + final++; + } + } + + ASSERT_EQ(0, final); +} +TEST_HARNESS_MAIN -- 2.17.1

2 years, 3 months

9
18
0 0

[PATCH bpf-next 0/2] update 32-bit bounds when the lower 32-bit value is not wrapping

by Xu Kuohai

This patchset updates __reg_combine_64_into_32 function to set 32-bit bounds when lower 32-bit value is not wrapping, and add cases to for it. Xu Kuohai (2): bpf: update 32-bit bounds when the lower 32-bit value is not wrapping selftests/bpf: check bounds not in the 32-bit range kernel/bpf/verifier.c | 27 ++-- tools/testing/selftests/bpf/verifier/bounds.c | 121 ++++++++++++++++++ 2 files changed, 132 insertions(+), 16 deletions(-) -- 2.30.2

2 years, 3 months

3
5
0 0

[PATCH 0/2] kselftest/alsa: Log card and control information during startup

by Mark Brown

These patches help make the logs a bit more friendly to work with by adding human readable names for cards and controls alongside the numbers assigned to them even when things are working well. Signed-off-by: Mark Brown <broonie(a)kernel.org> --- Mark Brown (2): kselftest/alsa - mixer: Always log control names kselftest/alsa: Log card names during startup tools/testing/selftests/alsa/mixer-test.c | 13 +++++++++++++ tools/testing/selftests/alsa/pcm-test.c | 10 ++++++++++ 2 files changed, 23 insertions(+) --- base-commit: fe15c26ee26efa11741a7b632e9f23b01aca4cc6 change-id: 20230223-alsa-log-ctl-name-fb07f30d7217 Best regards, -- Mark Brown <broonie(a)kernel.org>

2 years, 3 months

2
3
0 0

[PATCH] kselftest/alsa - mixer-test: Don't fail tests if we can't restore default

by Mark Brown

If a control has an invalid default value then we might fail to set it when restoring the default value after our write tests, for example due to correctly implemented range checks in put() operations. Currently this causes us to report the tests we were running as failed even when the operation we were trying to test is successful, making it look like there are problems where none really exist. Stop doing this, only reporting any issues during the actual test. We already have validation for the initial readback being in spec and for writing the default value back so failed tests will be reported for these controls, and we log an error on the operation that failed when we write so there will be a diagnostic warning the user that there is a problem. Signed-off-by: Mark Brown <broonie(a)kernel.org> --- tools/testing/selftests/alsa/mixer-test.c | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-) diff --git a/tools/testing/selftests/alsa/mixer-test.c b/tools/testing/selftests/alsa/mixer-test.c index 05f1749ae19d..ac5efa42d488 100644 --- a/tools/testing/selftests/alsa/mixer-test.c +++ b/tools/testing/selftests/alsa/mixer-test.c @@ -755,7 +755,6 @@ static bool test_ctl_write_valid_enumerated(struct ctl_data *ctl) static void test_ctl_write_valid(struct ctl_data *ctl) { bool pass; - int err; /* If the control is turned off let's be polite */ if (snd_ctl_elem_info_is_inactive(ctl->info)) { @@ -797,9 +796,7 @@ static void test_ctl_write_valid(struct ctl_data *ctl) } /* Restore the default value to minimise disruption */ - err = write_and_verify(ctl, ctl->def_val, NULL); - if (err < 0) - pass = false; + write_and_verify(ctl, ctl->def_val, NULL); ksft_test_result(pass, "write_valid.%d.%d\n", ctl->card->card, ctl->elem); @@ -1015,9 +1012,7 @@ static void test_ctl_write_invalid(struct ctl_data *ctl) } /* Restore the default value to minimise disruption */ - err = write_and_verify(ctl, ctl->def_val, NULL); - if (err < 0) - pass = false; + write_and_verify(ctl, ctl->def_val, NULL); ksft_test_result(pass, "write_invalid.%d.%d\n", ctl->card->card, ctl->elem); --- base-commit: fe15c26ee26efa11741a7b632e9f23b01aca4cc6 change-id: 20230224-alsa-mixer-test-restore-invalid-2a57b98aeb7f Best regards, -- Mark Brown <broonie(a)kernel.org>

2 years, 3 months

2
1
0 0

[PATCH v3 1/3] kunit: fix bug in debugfs logs of parameterized tests

by Rae Moar

Fix bug in debugfs logs that causes individual parameterized results to not appear because the log is reinitialized (cleared) when each parameter is run. Ensure these results appear in the debugfs logs, increase log size to allow for the size of parameterized results. As a result, append lines to the log directly rather than using an intermediate variable that can cause stack size warnings due to the increased log size. Here is the debugfs log of ext4_inode_test which uses parameterized tests before the fix: KTAP version 1 # Subtest: ext4_inode_test 1..1 # Totals: pass:16 fail:0 skip:0 total:16 ok 1 ext4_inode_test As you can see, this log does not include any of the individual parametrized results. After (in combination with the next two fixes to remove extra empty line and ensure KTAP valid format): KTAP version 1 1..1 KTAP version 1 # Subtest: ext4_inode_test 1..1 KTAP version 1 # Subtest: inode_test_xtimestamp_decoding ok 1 1901-12-13 Lower bound of 32bit < 0 timestamp, no extra bits ... (the rest of the individual parameterized tests) ok 16 2446-05-10 Upper bound of 32bit >=0 timestamp. All extra # inode_test_xtimestamp_decoding: pass:16 fail:0 skip:0 total:16 ok 1 inode_test_xtimestamp_decoding # Totals: pass:16 fail:0 skip:0 total:16 ok 1 ext4_inode_test Signed-off-by: Rae Moar <rmoar(a)google.com> Reviewed-by: David Gow <davidgow(a)google.com> --- Changes from v2 -> v3: - Fix a off-by-one bug in the kunit_log_append method. Changes from v1 -> v2: - Remove the use of the line variable in kunit_log_append that was causing stack size warnings. - Add before and after to the commit message. include/kunit/test.h | 2 +- lib/kunit/test.c | 18 ++++++++++++------ 2 files changed, 13 insertions(+), 7 deletions(-) diff --git a/include/kunit/test.h b/include/kunit/test.h index 08d3559dd703..0668d29f3453 100644 --- a/include/kunit/test.h +++ b/include/kunit/test.h @@ -34,7 +34,7 @@ DECLARE_STATIC_KEY_FALSE(kunit_running); struct kunit; /* Size of log associated with test. */ -#define KUNIT_LOG_SIZE 512 +#define KUNIT_LOG_SIZE 1500 /* Maximum size of parameter description string. */ #define KUNIT_PARAM_DESC_SIZE 128 diff --git a/lib/kunit/test.c b/lib/kunit/test.c index c9e15bb60058..c4d6304edd61 100644 --- a/lib/kunit/test.c +++ b/lib/kunit/test.c @@ -114,22 +114,27 @@ static void kunit_print_test_stats(struct kunit *test, */ void kunit_log_append(char *log, const char *fmt, ...) { - char line[KUNIT_LOG_SIZE]; va_list args; - int len_left; + int len, log_len, len_left; if (!log) return; - len_left = KUNIT_LOG_SIZE - strlen(log) - 1; + log_len = strlen(log); + len_left = KUNIT_LOG_SIZE - log_len - 1; if (len_left <= 0) return; + /* Evaluate length of line to add to log */ va_start(args, fmt); - vsnprintf(line, sizeof(line), fmt, args); + len = vsnprintf(NULL, 0, fmt, args) + 1; + va_end(args); + + /* Print formatted line to the log */ + va_start(args, fmt); + vsnprintf(log + log_len, min(len, len_left), fmt, args); va_end(args); - strncat(log, line, len_left); } EXPORT_SYMBOL_GPL(kunit_log_append); @@ -437,7 +442,6 @@ static void kunit_run_case_catch_errors(struct kunit_suite *suite, struct kunit_try_catch_context context; struct kunit_try_catch *try_catch; - kunit_init_test(test, test_case->name, test_case->log); try_catch = &test->try_catch; kunit_try_catch_init(try_catch, @@ -533,6 +537,8 @@ int kunit_run_tests(struct kunit_suite *suite) struct kunit_result_stats param_stats = { 0 }; test_case->status = KUNIT_SKIPPED; + kunit_init_test(&test, test_case->name, test_case->log); + if (!test_case->generate_params) { /* Non-parameterised test. */ kunit_run_case_catch_errors(suite, test_case, &test); base-commit: 60684c2bd35064043360e6f716d1b7c20e967b7d -- 2.40.0.rc0.216.gc4246ad0f0-goog

2 years, 3 months

2
5
0 0

Re: [PATCH bpf-next v2 0/8] Support defragmenting IPv(4|6) packets in BPF

by Alexei Starovoitov

On Mon, Feb 27, 2023 at 5:57 PM Daniel Xu <dxu(a)dxuuu.xyz> wrote: > > Hi Alexei, > > On Mon, Feb 27, 2023 at 03:03:38PM -0800, Alexei Starovoitov wrote: > > On Mon, Feb 27, 2023 at 12:51:02PM -0700, Daniel Xu wrote: > > > === Context === > > > > > > In the context of a middlebox, fragmented packets are tricky to handle. > > > The full 5-tuple of a packet is often only available in the first > > > fragment which makes enforcing consistent policy difficult. There are > > > really only two stateless options, neither of which are very nice: > > > > > > 1. Enforce policy on first fragment and accept all subsequent fragments. > > > This works but may let in certain attacks or allow data exfiltration. > > > > > > 2. Enforce policy on first fragment and drop all subsequent fragments. > > > This does not really work b/c some protocols may rely on > > > fragmentation. For example, DNS may rely on oversized UDP packets for > > > large responses. > > > > > > So stateful tracking is the only sane option. RFC 8900 [0] calls this > > > out as well in section 6.3: > > > > > > Middleboxes [...] should process IP fragments in a manner that is > > > consistent with [RFC0791] and [RFC8200]. In many cases, middleboxes > > > must maintain state in order to achieve this goal. > > > > > > === BPF related bits === > > > > > > However, when policy is enforced through BPF, the prog is run before the > > > kernel reassembles fragmented packets. This leaves BPF developers in a > > > awkward place: implement reassembly (possibly poorly) or use a stateless > > > method as described above. > > > > > > Fortunately, the kernel has robust support for fragmented IP packets. > > > This patchset wraps the existing defragmentation facilities in kfuncs so > > > that BPF progs running on middleboxes can reassemble fragmented packets > > > before applying policy. > > > > > > === Patchset details === > > > > > > This patchset is (hopefully) relatively straightforward from BPF perspective. > > > One thing I'd like to call out is the skb_copy()ing of the prog skb. I > > > did this to maintain the invariant that the ctx remains valid after prog > > > has run. This is relevant b/c ip_defrag() and ip_check_defrag() may > > > consume the skb if the skb is a fragment. > > > > Instead of doing all that with extra skb copy can you hook bpf prog after > > the networking stack already handled ip defrag? > > What kind of middle box are you doing? Why does it have to run at TC layer? > > Unless I'm missing something, the only other relevant hooks would be > socket hooks, right? > > Unfortunately I don't think my use case can do that. We are running the > kernel as a router, so no sockets are involved. Are you using bpf_fib_lookup and populating kernel routing table and doing everything on your own including neigh ? Have you considered to skb redirect to another netdev that does ip defrag? Like macvlan does it under some conditions. This can be generalized. Recently Florian proposed to allow calling bpf progs from all existing netfilter hooks. You can pretend to local deliver and hook in NF_INET_LOCAL_IN ? I feel it would be so much cleaner if stack does ip_defrag normally. The general issue of skb ownership between bpf prog and defrag logic isn't really solved with skb_copy. It's still an issue.

2 years, 3 months

4
6
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-kselftest-mirror