September 2025 - Linux-kselftest-mirror

[PATCH net-next v6] selftests: net: add test for ipv6 fragmentation

by Brett A C Sheffield

Add selftest for the IPv6 fragmentation regression which affected several stable kernels. Commit a18dfa9925b9 ("ipv6: save dontfrag in cork") was backported to stable without some prerequisite commits. This caused a regression when sending IPv6 UDP packets by preventing fragmentation and instead returning -1 (EMSGSIZE). Add selftest to check for this issue by attempting to send a packet larger than the interface MTU. The packet will be fragmented on a working kernel, with sendmsg(2) correctly returning the expected number of bytes sent. When the regression is present, sendmsg returns -1 and sets errno to EMSGSIZE. Link: https://lore.kernel.org/stable/aElivdUXqd1OqgMY@karahi.gladserv.com Signed-off-by: Brett A C Sheffield <bacs(a)librecast.net> Reviewed-by: Willem de Bruijn <willemb(a)google.com> --- Thanks again Willem for the review. v6 changes: - port -> network byte order - remove unecessary variable err - check return value of close() v5 changes: - disable_dad: delete - not needed for lo - main: simplify failure paths - main: char -> static char buf - setup: remove pointless return value - setup: remove unused variable fd - setup: merge with interface_up() to simplify - setup: check all system call return values - remove no longer used headers v4 changes: - fix "else should follow close brace" (checkpatch ERROR) v3 changes: - add usleep instead of busy polling on sendmsg - simplify error handling by using error() and leaving cleanup to O/S - use loopback interface - don't bother creating TAP - send to localhost (::1) v2 changes: - remove superfluous namespace calls - unshare(2) suffices - remove usleep(). Don't wait for the interface to be ready, just send, and handle the (less likely) error case by retrying. - set destination address only once - document our use of the IPv6 link-local source address - send to port 9 (DISCARD) instead of 4242 (DONT PANIC) - ensure sockets are closed on failure paths - use KSFT exit codes for clarity v5: https://lore.kernel.org/netdev/20250902142502.27278-1-bacs@librecast.net v4: https://lore.kernel.org/netdev/20250901123757.13112-1-bacs@librecast.net v3: https://lore.kernel.org/netdev/20250901112248.5218-1-bacs@librecast.net v2: https://lore.kernel.org/netdev/20250831102908.14655-1-bacs@librecast.net v1: https://lore.kernel.org/netdev/20250825092548.4436-3-bacs@librecast.net tools/testing/selftests/net/.gitignore | 1 + tools/testing/selftests/net/Makefile | 1 + .../selftests/net/ipv6_fragmentation.c | 114 ++++++++++++++++++ 3 files changed, 116 insertions(+) create mode 100644 tools/testing/selftests/net/ipv6_fragmentation.c diff --git a/tools/testing/selftests/net/.gitignore b/tools/testing/selftests/net/.gitignore index 47c293c2962f..3d4b4a53dfda 100644 --- a/tools/testing/selftests/net/.gitignore +++ b/tools/testing/selftests/net/.gitignore @@ -16,6 +16,7 @@ ip_local_port_range ipsec ipv6_flowlabel ipv6_flowlabel_mgr +ipv6_fragmentation log.txt msg_oob msg_zerocopy diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index eef0b8f8a7b0..276e0481d996 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -117,6 +117,7 @@ TEST_GEN_FILES += tfo TEST_PROGS += tfo_passive.sh TEST_PROGS += broadcast_pmtu.sh TEST_PROGS += ipv6_force_forwarding.sh +TEST_GEN_PROGS += ipv6_fragmentation TEST_PROGS += route_hint.sh # YNL files, must be before "include ..lib.mk" diff --git a/tools/testing/selftests/net/ipv6_fragmentation.c b/tools/testing/selftests/net/ipv6_fragmentation.c new file mode 100644 index 000000000000..267ef62b5c72 --- /dev/null +++ b/tools/testing/selftests/net/ipv6_fragmentation.c @@ -0,0 +1,114 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Author: Brett A C Sheffield <bacs(a)librecast.net> + * + * Kernel selftest for the IPv6 fragmentation regression which affected stable + * kernels: + * + * https://lore.kernel.org/stable/aElivdUXqd1OqgMY@karahi.gladserv.com + * + * Commit: a18dfa9925b9 ("ipv6: save dontfrag in cork") was backported to stable + * without some prerequisite commits. + * + * This caused a regression when sending IPv6 UDP packets by preventing + * fragmentation and instead returning -1 (EMSGSIZE). + * + * This selftest demonstrates the issue by sending an IPv6 UDP packet to + * localhost (::1) on the loopback interface from the autoconfigured link-local + * address. + * + * sendmsg(2) returns bytes sent correctly on a working kernel, and returns -1 + * (EMSGSIZE) when the regression is present. + * + * The regression was not present in the mainline kernel, but add this test to + * catch similar breakage in future. + */ + +#define _GNU_SOURCE + +#include <error.h> +#include <net/if.h> +#include <netinet/in.h> +#include <sched.h> +#include <stdio.h> +#include <sys/ioctl.h> +#include <sys/socket.h> +#include <unistd.h> +#include "../kselftest.h" + +#define MTU 1500 +#define LARGER_THAN_MTU 8192 + +static void setup(void) +{ + struct ifreq ifr = { + .ifr_name = "lo" + }; + int ctl; + + /* we need to set MTU, so do this in a namespace to play nicely */ + if (unshare(CLONE_NEWNET) == -1) + error(KSFT_FAIL, errno, "unshare"); + + ctl = socket(AF_LOCAL, SOCK_STREAM, 0); + if (ctl == -1) + error(KSFT_FAIL, errno, "socket"); + + /* ensure MTU is smaller than what we plan to send */ + ifr.ifr_mtu = MTU; + if (ioctl(ctl, SIOCSIFMTU, &ifr) == -1) + error(KSFT_FAIL, errno, "ioctl: set MTU"); + + /* bring up interface */ + if (ioctl(ctl, SIOCGIFFLAGS, &ifr) == -1) + error(KSFT_FAIL, errno, "ioctl SIOCGIFFLAGS"); + ifr.ifr_flags = ifr.ifr_flags | IFF_UP; + if (ioctl(ctl, SIOCSIFFLAGS, &ifr) == -1) + error(KSFT_FAIL, errno, "ioctl: bring interface up"); + + if (close(ctl) == -1) + error(KSFT_FAIL, errno, "close"); +} + +int main(void) +{ + struct in6_addr addr = { + .s6_addr[15] = 0x01, /* ::1 */ + }; + struct sockaddr_in6 sa = { + .sin6_family = AF_INET6, + .sin6_addr = addr, + .sin6_port = htons(9) /* port 9/udp (DISCARD) */ + }; + static char buf[LARGER_THAN_MTU] = {0}; + struct iovec iov = { .iov_base = buf, .iov_len = sizeof(buf) }; + struct msghdr msg = { + .msg_iov = &iov, + .msg_iovlen = 1, + .msg_name = (struct sockaddr *)&sa, + .msg_namelen = sizeof(sa), + }; + ssize_t rc; + int s; + + printf("Testing IPv6 fragmentation\n"); + setup(); + s = socket(AF_INET6, SOCK_DGRAM, 0); +send_again: + rc = sendmsg(s, &msg, 0); + if (rc == -1) { + /* if interface wasn't ready, try again */ + if (errno == EADDRNOTAVAIL) { + usleep(1000); + goto send_again; + } + error(KSFT_FAIL, errno, "sendmsg"); + } else if (rc != LARGER_THAN_MTU) { + error(KSFT_FAIL, errno, "sendmsg returned %zi, expected %i", + rc, LARGER_THAN_MTU); + } + printf("[PASS] sendmsg() returned %zi\n", rc); + if (close(s) == -1) + error(KSFT_FAIL, errno, "close"); + return KSFT_PASS; +} base-commit: cd8a4cfa6bb43a441901e82f5c222dddc75a18a3 -- 2.49.1

1 week, 5 days

2
1
0 0

[PATCHv5 net-next 0/3] bonding: support aggregator selection based on port priority

by Hangbin Liu

This patchset introduces a new per-port bonding option: `ad_actor_port_prio`. It allows users to configure the actor's port priority, which can then be used by the bonding driver for aggregator selection based on port priority. This provides finer control over LACP aggregator choice, especially in setups with multiple eligible aggregators over 2 switches. v5: a) rename 'prio' to 'actor_port_prio' in bond_ad_select_tbl (Jay Vosburgh) b) update document description v4: a) fix actor_port_prio minimal value (Jay Vosburgh) b) fix ad_agg_selection_test comment order (Paolo Abeni) c) restruct selftest, reduce duplication (Paolo Abeni) v3: a) add comments when init slave port_priority (Jonas Gorski) b) rename ad_lacp_port_prio to lacp_port_prio (Jay Vosburgh) v2: a) set default bond option value for port priority (Nikolay Aleksandrov) b) fix __agg_ports_priority coding style (Nikolay Aleksandrov) c) fix shellcheck warns Hangbin Liu (3): bonding: add support for per-port LACP actor priority bonding: support aggregator selection based on port priority selftests: bonding: add test for LACP actor port priority Documentation/networking/bonding.rst | 25 +++- drivers/net/bonding/bond_3ad.c | 31 +++++ drivers/net/bonding/bond_netlink.c | 16 +++ drivers/net/bonding/bond_options.c | 45 +++++++- include/net/bond_3ad.h | 2 + include/net/bond_options.h | 1 + include/uapi/linux/if_link.h | 1 + .../selftests/drivers/net/bonding/Makefile | 3 +- .../drivers/net/bonding/bond_lacp_prio.sh | 108 ++++++++++++++++++ tools/testing/selftests/net/forwarding/lib.sh | 24 ---- tools/testing/selftests/net/lib.sh | 24 ++++ 11 files changed, 247 insertions(+), 33 deletions(-) create mode 100755 tools/testing/selftests/drivers/net/bonding/bond_lacp_prio.sh -- 2.50.1

1 week, 5 days

2
4
0 0

[PATCH] selftests/kselftest_harness: Add harness-selftest.expected to TEST_FILES

by Yi Lai

The harness-selftest.expected is not installed in INSTALL_PATH. Attempting to execute harness-selftest.sh shows warning: diff: ./kselftest_harness/harness-selftest.expected: No such file or directory Add harness-selftest.expected to TEST_FILES. Signed-off-by: Yi Lai <yi1.lai(a)intel.com> --- tools/testing/selftests/kselftest_harness/Makefile | 1 + 1 file changed, 1 insertion(+) diff --git a/tools/testing/selftests/kselftest_harness/Makefile b/tools/testing/selftests/kselftest_harness/Makefile index 0617535a6ce4..d2369c01701a 100644 --- a/tools/testing/selftests/kselftest_harness/Makefile +++ b/tools/testing/selftests/kselftest_harness/Makefile @@ -2,6 +2,7 @@ TEST_GEN_PROGS_EXTENDED := harness-selftest TEST_PROGS := harness-selftest.sh +TEST_FILES := harness-selftest.expected EXTRA_CLEAN := harness-selftest.seen include ../lib.mk -- 2.43.0

1 week, 6 days

4
3
0 0

[PATCH v2 0/3] Fix va_high_addr_switch.sh test failure

by Chunyu Hu

The three patches fix the va_high_addr_switch.sh test failure on x86_64. Patch 1 fixes the hugepage setup issue that nr_hugepages is reset too early in run_vmtests.sh and break the later va_high_addr_switch testing. Patch 2 adds hugepage setup in va_high_addr_switch test, so that it can still work if vm_runtests.sh changes the hugepage setup someday. Patch 3 fixes the test failure caused by the hint addr align method change in hugetlb_get_unmapped_area(). Changes in v2: - patch 1 renames nr_hugepgs_origin to orig_nr_hugepgs - add a patch 2 to setup hugeapges in va_high_addr_switch test Chunyu Hu (3): selftests/mm: fix hugepages cleanup too early selftests/mm: alloc hugepages in va_high_addr_switch test selftests/mm: fix va_high_addr_switch.sh failure on x86_64 tools/testing/selftests/mm/run_vmtests.sh | 9 ++++- .../selftests/mm/va_high_addr_switch.c | 4 +- .../selftests/mm/va_high_addr_switch.sh | 37 +++++++++++++++++++ 3 files changed, 46 insertions(+), 4 deletions(-) -- 2.49.0

1 week, 6 days

2
6
0 0

[PATCH v2 0/2] selftests/mm: uffd-stress fixes

by Dev Jain

This patchset ensures that the number of hugepages is correctly set in the system so that the uffd-stress test does not fail due to the racy nature of the test. Patch 1 changes the hugepage constraint in the run_vmtests.sh script, whereas patch 2 changes the constraint in the test itself. --- Based on 6.17-rc5. Dev Jain (2): selftests/mm/uffd-stress: make test operate on less hugetlb memory selftests/mm/uffd-stress: stricten constraint on free hugepages needed before the test tools/testing/selftests/mm/run_vmtests.sh | 10 +++++++--- tools/testing/selftests/mm/uffd-stress.c | 17 +++++++++++------ 2 files changed, 18 insertions(+), 9 deletions(-) -- 2.30.2

1 week, 6 days

1
3
0 0

[PATCH 0/2] selftests/mm: uffd-stress fixes

by Dev Jain

This patchset ensures that the number of hugepages is correctly set in the system so that the uffd-stress test does not fail due to the racy nature of the test. Patch 1 corrects the hugepage constraint in the run_vmtests.sh script, whereas patch 2 corrects the constraint in the test itself. Dev Jain (2): selftests/mm/uffd-stress: Make test operate on less hugetlb memory selftests/mm/uffd-stress: Stricten constraint on free hugepages before the test tools/testing/selftests/mm/run_vmtests.sh | 2 +- tools/testing/selftests/mm/uffd-stress.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) -- 2.30.2

1 week, 6 days

3
7
0 0

[PATCHv2 net-next 0/5] net: common feature compute for upper interface

by Hangbin Liu

Some high-level virtual drivers need to compute features from their lower devices, but each currently has its own implementation and may miss some feature computations. This patch set introduces a common function to compute features for such devices. Currently, bonding, team, and bridge have been updated to use the new helper. v2: a) remove hard_header_len setting. I will set needed_headroom for bond/team in a separate patch as bridge has it's own ways. (Ido Schimmel) b) Add test file to Makefile, set RET=0 to a proper location. (Ido Schimmel) Hangbin Liu (5): net: add a common function to compute features from lowers devices bonding: use common function to compute the features team: use common function to compute the features net: bridge: use common function to compute the features selftests/net: add offload checking test for virtual interface drivers/net/bonding/bond_main.c | 99 +---------- drivers/net/team/team_core.c | 73 +------- include/linux/netdevice.h | 19 +++ net/bridge/br_if.c | 22 +-- net/core/dev.c | 76 +++++++++ tools/testing/selftests/net/Makefile | 1 + tools/testing/selftests/net/config | 2 + tools/testing/selftests/net/vdev_offload.sh | 176 ++++++++++++++++++++ 8 files changed, 285 insertions(+), 183 deletions(-) create mode 100755 tools/testing/selftests/net/vdev_offload.sh -- 2.50.1

1 week, 6 days

3
14
0 0

[PATCH] selftests/bpf: Fix incorrect array size calculation

by Jiayuan Chen

Fix it by using ARRAY_SIZE. Signed-off-by: Jiayuan Chen <jiayuan.chen(a)linux.dev> Reported-by: Dan Carpenter <dan.carpenter(a)linaro.org> Closes: https://lore.kernel.org/bpf/aLqfWuRR9R_KTe5e@stanley.mountain/ --- tools/testing/selftests/bpf/benchs/bench_sockmap.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/tools/testing/selftests/bpf/benchs/bench_sockmap.c b/tools/testing/selftests/bpf/benchs/bench_sockmap.c index 8ebf563a67a2..cfc072aa7fff 100644 --- a/tools/testing/selftests/bpf/benchs/bench_sockmap.c +++ b/tools/testing/selftests/bpf/benchs/bench_sockmap.c @@ -10,6 +10,7 @@ #include <argp.h> #include "bench.h" #include "bench_sockmap_prog.skel.h" +#include "bpf_util.h" #define FILE_SIZE (128 * 1024) #define DATA_REPEAT_SIZE 10 @@ -124,8 +125,8 @@ static void bench_sockmap_prog_destroy(void) { int i; - for (i = 0; i < sizeof(ctx.fds); i++) { - if (ctx.fds[0] > 0) + for (i = 0; i < ARRAY_SIZE(ctx.fds); i++) { + if (ctx.fds[i] > 0) close(ctx.fds[i]); } base-commit: 60ef54156148ffaa719845bc5b1fdeafa67763fc -- 2.43.0

1 week, 6 days

2
1
0 0

[PATCH 0/5] KVM: Improve nested VMX performance

by Fred Griffoul

From: Fred Griffoul <griffoul(a)casper.infradead.org> This patch series addresses performance issues in nested VMX when handling unmanaged guest memory. Unmanaged guest memory refers to memory not directly mapped by the kernel (no struct page), such as memory passed with the mem= parameter or guest_memfd for non-Confidential Computing (CoCo) VMs. Current Problem: During nested VMX operations, the system frequently accesses specific guest pages during L2 VM entry/exit cycles. The current workflow: 1. kvm_vcpu_map() invokes memremap() for unmanaged memory. 2. The system either directly accesses mapped memory via nested VMX or passes it to the L2 guest through vmcs02. 3. kvm_vcpu_unmap() invokes memunmap() This repeated map/unmap cycle creates significant performance overhead due to expensive remapping operations. Solution approach: Our solution replaces kvm_host_map with gfn_to_pfn_cache in nested VMX. It addresses two distinct types of guest pages. First, we handle the L1 MSR bitmap page, which requires read-only access for folding L1 and L0 MSR bitmap. We implement this conversion to gfn_to_pfn_cache in patch 1. Second, we tackle system pages, including APIC access, virtual APIC, and posted interrupt descriptor pages. These pages are more complex as they're accessed by both nested VMX code _and_ passed to the L2 guest in vmcs02 fields. This requires to restore and complete the "guest-uses-pfn" support in pfncache through patches 2 and 3, followed by implementing kvm_host_map replacement with caches in patch 4. Testing: Patch 5 introduces a new selftest to verify cache invalidation and memslot update functionality. The changes are available in a git repository at: git://git.infradead.org/users/griffoul/linux.git tags/nvmx-gpc-v1 Suggested-by: dwmw(a)amazon.co.uk Fred Griffoul (5): KVM: nVMX: Implement cache for L1 MSR bitmap KVM: pfncache: Restore guest-uses-pfn support KVM: x86: Add nested state validation for pfncache support KVM: nVMX: Implement cache for L1 APIC pages KVM: selftests: Add nested VMX APIC cache invalidation test arch/x86/include/asm/kvm_host.h | 1 + arch/x86/kvm/vmx/nested.c | 213 +++++++++--- arch/x86/kvm/vmx/vmx.h | 10 +- arch/x86/kvm/x86.c | 14 +- include/linux/kvm_host.h | 34 +- include/linux/kvm_types.h | 1 + tools/testing/selftests/kvm/Makefile.kvm | 1 + .../selftests/kvm/x86/vmx_apic_update_test.c | 302 ++++++++++++++++++ virt/kvm/kvm_main.c | 3 +- virt/kvm/kvm_mm.h | 6 +- virt/kvm/pfncache.c | 43 ++- 11 files changed, 575 insertions(+), 53 deletions(-) create mode 100644 tools/testing/selftests/kvm/x86/vmx_apic_update_test.c -- 2.51.0

1 week, 6 days

1
5
0 0

[PATCH] selftests/kvm: remove stale TODO in xapic_state_test

by Sukrut Heroorkar

The TODO about using the number of vCPUs instead of vcpu.id + 1 was already addressed by commit 376bc1b458c9 ("KVM: selftests: Don't assume vcpu->id is '0' in xAPIC state test"). The comment is now stale and can be removed. Signed-off-by: Sukrut Heroorkar <hsukrut3(a)gmail.com> --- tools/testing/selftests/kvm/x86/xapic_state_test.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tools/testing/selftests/kvm/x86/xapic_state_test.c b/tools/testing/selftests/kvm/x86/xapic_state_test.c index fdebff1165c7..3b4814c55722 100644 --- a/tools/testing/selftests/kvm/x86/xapic_state_test.c +++ b/tools/testing/selftests/kvm/x86/xapic_state_test.c @@ -120,8 +120,8 @@ static void test_icr(struct xapic_vcpu *x) __test_icr(x, icr | i); /* - * Send all flavors of IPIs to non-existent vCPUs. TODO: use number of - * vCPUs, not vcpu.id + 1. Arbitrarily use vector 0xff. + * Send all flavors of IPIs to non-existent vCPUs. Arbitrarily use + * vector 0xff. */ icr = APIC_INT_ASSERT | 0xff; for (i = 0; i < 0xff; i++) { -- 2.43.0

1 week, 6 days

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-kselftest-mirror September 2025