- 13 Jul, 2020 16 commits
-
-
Alan Maguire authored
The bpf helper bpf_trace_printk() uses trace_printk() under the hood. This leads to an alarming warning message originating from trace buffer allocation which occurs the first time a program using bpf_trace_printk() is loaded. We can instead create a trace event for bpf_trace_printk() and enable it in-kernel when/if we encounter a program using the bpf_trace_printk() helper. With this approach, trace_printk() is not used directly and no warning message appears. This work was started by Steven (see Link) and finished by Alan; added Steven's Signed-off-by with his permission. Signed-off-by:
Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by:
Alan Maguire <alan.maguire@oracle.com> Signed-off-by:
Alexei Starovoitov <ast@kernel.org> Acked-by:
Andrii Nakryiko <andriin@fb.com> Link: https://lore.kernel.org/r/20200628194334.6238b933@oasis.local.home Link: https://lore.kernel.org/bpf/1594641154-18897-2-git-send-email-alan.maguire@oracle.com
-
Andrii Nakryiko authored
Don't emit warning that bpftool was built without PID iterator support. This error garbles JSON output of otherwise perfectly valid show commands. Reported-by:
Andrey Ignatov <rdna@fb.com> Signed-off-by:
Yonghong Song <yhs@fb.com> Link: https://lore.kernel.org/bpf/20200710232605.20918-1-andriin@fb.com
-
Alexei Starovoitov authored
Ciara Loftus says: ==================== This series introduces new statistics for af_xdp: 1. drops due to rx ring being full 2. drops due to fill ring being empty 3. failures pulling an item from the tx ring These statistics should assist users debugging and troubleshooting peformance issues and packet drops. The statistics are made available though the getsockopt and xsk_diag interfaces, and the ability to dump these extended statistics is made available in the xdpsock application via the --extra-stats or -x flag. A separate patch which will add ss/iproute2 support will follow. ==================== Acked-by:
Björn Töpel <bjorn.topel@intel.com> Signed-off-by:
-
Ciara Loftus authored
Add xdp statistics to the information dumped through the xsk_diag interface Signed-off-by:
Ciara Loftus <ciara.loftus@intel.com> Signed-off-by:
-
Ciara Loftus authored
Introduce the --extra-stats (or simply -x) flag to the xdpsock application which prints additional statistics alongside the regular rx and tx counters. The new statistics printed report error conditions eg. rx ring full, invalid descriptors, etc. Signed-off-by:
-
Ciara Loftus authored
It can be useful for the user to know the reason behind a dropped packet. Introduce new counters which track drops on the receive path caused by: 1. rx ring being full 2. fill ring being empty Also, on the tx path introduce a counter which tracks the number of times we attempt pull from the tx ring when it is empty. Signed-off-by:
-
Alexei Starovoitov authored
Jiri Olsa says: ==================== This patchset adds: - support to generate BTF ID lists that are resolved during kernel linking and usable within kernel code with following macros: BTF_ID_LIST(bpf_skb_output_btf_ids) BTF_ID(struct, sk_buff) and access it in kernel code via: extern u32 bpf_skb_output_btf_ids[]; - resolve_btfids tool that scans elf object for .BTF_ids section and resolves its symbols with BTF ID values - resolving of bpf_ctx_convert struct and several other objects with BTF_ID_LIST v7 changes: - added more acks [Andrii] - added some name-conflicting entries and fixed resolve_btfids to process them properly [Andrii] - changed bpf_get_task_stack_proto to use BTF_IDS_LIST/BTF_ID macros [Andrii] - fixed selftest build for resolve_btfids test ==================== Signed-off-by: -
Jiri Olsa authored
Adding resolve_btfids test under test_progs suite. It's possible to use btf_ids.h header and its logic in user space application, so we can add easy test for it. The test defines BTF_ID_LIST and checks it gets properly resolved. For this reason the test_progs binary (and other binaries that use TRUNNER* macros) is processed with resolve_btfids tool, which resolves BTF IDs in .BTF_ids section. The BTF data are taken from btf_data.o object rceated from progs/btf_data.c. Signed-off-by:
Jiri Olsa <jolsa@kernel.org> Signed-off-by:
-
Jiri Olsa authored
It will be needed by bpf selftest for resolve_btfids tool. Also adding __PASTE macro as btf_ids.h dependency, which is defined in: include/linux/compiler_types.h but because tools/include do not have this header, I'm putting the macro into linux/compiler.h header. Signed-off-by:
-
Jiri Olsa authored
Updating btf.rst doc with info about .BTF_ids section Signed-off-by:
-
Jiri Olsa authored
This way the ID is resolved during compile time, and we can remove the runtime name search. Signed-off-by:
-
Jiri Olsa authored
Now when we moved the helpers btf_id arrays into .BTF_ids section, we can remove the code that resolve those IDs in runtime. Signed-off-by:
-
Jiri Olsa authored
Using BTF_ID_LIST macro to define lists for several helpers using BTF arguments. And running resolve_btfids on vmlinux elf object during linking, so the .BTF_ids section gets the IDs resolved. Signed-off-by:
-
Jiri Olsa authored
Adding support to generate .BTF_ids section that will hold BTF ID lists for verifier. Adding macros that will help to define lists of BTF ID values placed in .BTF_ids section. They are initially filled with zeros (during compilation) and resolved later during the linking phase by resolve_btfids tool. Following defines list of one BTF ID value: BTF_ID_LIST(bpf_skb_output_btf_ids) BTF_ID(struct, sk_buff) It also defines following variable to access the list: extern u32 bpf_skb_output_btf_ids[]; The BTF_ID_UNUSED macro defines 4 zero bytes. It's used when we want to define 'unused' entry in BTF_ID_LIST, like: BTF_ID_LIST(bpf_skb_output_btf_ids) BTF_ID(struct, sk_buff) BTF_ID_UNUSED BTF_ID(struct, task_struct) Suggested-by:
-
Jiri Olsa authored
The resolve_btfids tool will be used during the vmlinux linking, so it's necessary it's ready for it. Signed-off-by:
-
Jiri Olsa authored
The resolve_btfids tool scans elf object for .BTF_ids section and resolves its symbols with BTF ID values. It will be used to during linking time to resolve arrays of BTF ID values used in verifier, so these IDs do not need to be resolved in runtime. The expected layout of .BTF_ids section is described in main.c header. Related kernel changes are coming in following changes. Build issue reported by 0-DAY CI Kernel Test Service. Signed-off-by:
-
- 10 Jul, 2020 2 commits
-
-
Wenbo Zhang authored
The `BPF_LOG_BUF_SIZE`'s value is `UINT32_MAX >> 8`, so define an array with it on stack caused an overflow. Signed-off-by:
Wenbo Zhang <ethercflow@gmail.com> Signed-off-by:
Daniel Borkmann <daniel@iogearbox.net> Acked-by:
-
Andrii Nakryiko authored
Coverity's static analysis helpfully reported a memory leak introduced by 0f0e55d8 ("libbpf: Improve BTF sanitization handling"). While fixing it, I realized that btf__new() already creates a memory copy, so there is no need to do this. So this patch also fixes misleading btf__new() signature to make data into a `const void *` input parameter. And it avoids unnecessary memory allocation and copy in BTF sanitization code altogether. Fixes: 0f0e55d8 ("libbpf: Improve BTF sanitization handling") Signed-off-by:
-
- 08 Jul, 2020 10 commits
-
-
Daniel Borkmann authored
Andrii Nakryiko says: ==================== This patch set improves libbpf's support of old kernels, missing features like BTF support, global variables support, etc. Most critical one is a silent drop of CO-RE relocations if libbpf fails to load BTF (despite sanitization efforts). This is frequently the case for kernels that have no BTF support whatsoever. There are still useful BPF applications that could work on such kernels and do rely on CO-RE. To that end, this series revamps the way BTF is handled in libbpf. Failure to load BTF into kernel doesn't prevent libbpf from using BTF in its full capability (e.g., for CO-RE relocations) internally. Another issue that was identified was reliance of perf_buffer__new() on BPF_OBJ_GET_INFO_BY_FD command, which is more recent that perf_buffer support itself. Furthermore, BPF_OBJ_GET_INFO_BY_FD is needed just for some sanity checks to provide better user errors, so could be safely omitted if kernel doesn't provide it. Perf_buffer selftest was adjusted to use skeleton, instead of bpf_prog_load(). The latter uses BPF_F_TEST_RND_HI32 flag, which is a relatively recent addition and unnecessary fails selftest in libbpf's Travis CI tests. By using skeleton we both get a shorter selftest and it work on pretty ancient kernels, giving better libbpf test coverage. One new selftest was added that relies on basic CO-RE features, but otherwise doesn't expect any recent features (like global variables) from kernel. Again, it's good to have better coverage of old kernels in libbpf testing. ==================== Signed-off-by: -
Andrii Nakryiko authored
Switch perf_buffer test to use skeleton to avoid use of bpf_prog_load() and make test a bit more succinct. Also switch BPF program to use tracepoint instead of kprobe, as that allows to support older kernels, which had tracepoint support before kprobe support in the form that libbpf expects (i.e., libbpf expects /sys/bus/event_source/devices/kprobe/type, which doesn't always exist on old kernels). Signed-off-by:
-
Andrii Nakryiko authored
perf_buffer__new() is relying on BPF_OBJ_GET_INFO_BY_FD availability for few sanity checks. OBJ_GET_INFO for maps is actually much more recent feature than perf_buffer support itself, so this causes unnecessary problems on old kernels before BPF_OBJ_GET_INFO_BY_FD was added. This patch makes those sanity checks optional and just assumes best if command is not supported. If user specified something incorrectly (e.g., wrong map type), kernel will reject it later anyway, except user won't get a nice explanation as to why it failed. This seems like a good trade off for supporting perf_buffer on old kernels. Signed-off-by:
-
Andrii Nakryiko authored
Add a test that relies on CO-RE, but doesn't expect any of the recent features, not available on old kernels. This is useful for Travis CI tests running against very old kernels (e.g., libbpf has 4.9 kernel testing now), to verify that CO-RE still works, even if kernel itself doesn't support BTF yet, as long as there is .BTF embedded into vmlinux image by pahole. Given most of CO-RE doesn't require any kernel awareness of BTF, it is a useful test to validate that libbpf's BTF sanitization is working well even with ancient kernels. Signed-off-by:
-
Andrii Nakryiko authored
Change sanitization process to preserve original BTF, which might be used by libbpf itself for Kconfig externs, CO-RE relocs, etc, even if kernel is old and doesn't support BTF. To achieve that, if libbpf detects the need for BTF sanitization, it would clone original BTF, sanitize it in-place, attempt to load it into kernel, and if successful, will preserve loaded BTF FD in original `struct btf`, while freeing sanitized local copy. If kernel doesn't support any BTF, original btf and btf_ext will still be preserved to be used later for CO-RE relocation and other BTF-dependent libbpf features, which don't dependon kernel BTF support. Patch takes care to not specify BTF and BTF.ext features when loading BPF programs and/or maps, if it was detected that kernel doesn't support BTF features. Signed-off-by:
-
Andrii Nakryiko authored
Add setter for BTF FD to allow application more fine-grained control in more advanced scenarios. Storing BTF FD inside `struct btf` provides little benefit and probably would be better done differently (e.g., btf__load() could just return FD on success), but we are stuck with this due to backwards compatibility. The main problem is that it's impossible to load BTF and than free user-space memory, but keep FD intact, because `struct btf` assumes ownership of that FD upon successful load and will attempt to close it during btf__free(). To allow callers (e.g., libbpf itself for BTF sanitization) to have more control over this, add btf__set_fd() to allow to reset FD arbitrarily, if necessary. Signed-off-by:
-
Andrii Nakryiko authored
With valid ELF and valid BTF, there is no reason (apart from bugs) why BTF finalization should fail. So make it strict and return error if it fails. This makes CO-RE relocation more reliable, as they are not going to be just silently skipped, if BTF finalization failed. Signed-off-by:
-
Jesper Dangaard Brouer authored
There are a number of places in test_progs that use minus-1 as the argument to exit(). This is confusing as a process exit status is masked to be a number between 0 and 255 as defined in man exit(3). Thus, users will see status 255 instead of minus-1. This patch use positive exit code 3 instead of minus-1. These cases are put in the same group of infrastructure setup errors. Fixes: fd27b183 ("selftests/bpf: Reset process and thread affinity after each test/sub-test") Fixes: 811d7e37 ("bpf: selftests: Restore netns after each test") Signed-off-by:
Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by:
-
Jesper Dangaard Brouer authored
This is a follow up adjustment to commit 6c92bd5c ("selftests/bpf: Test_progs indicate to shell on non-actions"), that returns shell exit indication EXIT_FAILURE (value 1) when user selects a non-existing test. The problem with using EXIT_FAILURE is that a shell script cannot tell the difference between a non-existing test and the test failing. This patch uses value 2 as shell exit indication. (Aside note unrecognized option parameters use value 64). Fixes: 6c92bd5c ("selftests/bpf: Test_progs indicate to shell on non-actions") Signed-off-by:
-
Louis Peens authored
emit_obj_refs_json needs to added the same as with emit_obj_refs_plain to prevent segfaults, similar to Commit "8ae4121b bpf: Fix bpftool without skeleton code enabled"). See the error below: # ./bpftool -p prog { "error": "bpftool built without PID iterator support" },[{ "id": 2, "type": "cgroup_skb", "tag": "7be49e3934a125ba", "gpl_compatible": true, "loaded_at": 1594052789, "uid": 0, "bytes_xlated": 296, "jited": true, "bytes_jited": 203, "bytes_memlock": 4096, "map_ids": [2,3 Segmentation fault (core dumped) The same happens for ./bpftool -p map, as well as ./bpftool -j prog/map. Fixes: d53dee3f ("tools/bpftool: Show info for processes holding BPF map/prog/link/btf FDs") Signed-off-by:
Louis Peens <louis.peens@netronome.com> Signed-off-by:
Simon Horman <simon.horman@netronome.com> Reviewed-by:
Quentin Monnet <quentin@isovalent.com> Link: https://lore.kernel.org/bpf/20200708110827.7673-1-louis.peens@netronome.com
-
- 07 Jul, 2020 12 commits
-
-
Daniel T. Lee authored
samples/bpf no longer use bpf_map_def_legacy and instead use the libbpf's bpf_map_def or new BTF-defined MAP format. This commit removes unused bpf_map_def_legacy struct from selftests/bpf/bpf_legacy.h. Signed-off-by:
Daniel T. Lee <danieltimlee@gmail.com> Signed-off-by:
-
Daniel T. Lee authored
Previously, in order to set the numa_node attribute at the time of map creation using "libbpf", it was necessary to call bpf_create_map_node() directly (bpf_load approach), instead of calling bpf_object_load() that handles everything on its own, including map creation. And because of this problem, this sample had problems with refactoring from bpf_load to libbbpf. However, by commit 1bdb6c9a ("libbpf: Add a bunch of attribute getters/setters for map definitions") added the numa_node attribute and allowed it to be set in the map. By using libbpf instead of bpf_load, the inner map definition has been explicitly declared with BTF-defined format. Also, the element of ARRAY_OF_MAPS was also statically specified using the BTF format. And for this reason some logic in fixup_map() was not needed and changed or removed. Signed-off-by:
-
Daniel T. Lee authored
From commit 646f02ff ("libbpf: Add BTF-defined map-in-map support"), a way to define internal map in BTF-defined map has been added. Instead of using previous 'inner_map_idx' definition, the structure to be used for the inner map can be directly defined using array directive. __array(values, struct inner_map) This commit refactors map in map test program with libbpf by explicitly defining inner map with BTF-defined format. Signed-off-by:
-
Daniel T. Lee authored
Currently, BPF programs with kprobe/sys_connect does not work properly. Commit 34745aed ("samples/bpf: fix kprobe attachment issue on x64") This commit modifies the bpf_load behavior of kprobe events in the x64 architecture. If the current kprobe event target starts with "sys_*", add the prefix "__x64_" to the front of the event. Appending "__x64_" prefix with kprobe/sys_* event was appropriate as a solution to most of the problems caused by the commit below. commit d5a00528 ("syscalls/core, syscalls/x86: Rename struct pt_regs-based sys_*() to __x64_sys_*()") However, there is a problem with the sys_connect kprobe event that does not work properly. For __sys_connect event, parameters can be fetched normally, but for __x64_sys_connect, parameters cannot be fetched. ffffffff818d3520 <__x64_sys_connect>: ffffffff818d3520: e8 fb df 32 00 callq 0xffffffff81c01520 <__fentry__> ffffffff818d3525: 48 8b 57 60 movq 96(%rdi), %rdx ffffffff818d3529: 48 8b 77 68 movq 104(%rdi), %rsi ffffffff818d352d: 48 8b 7f 70 movq 112(%rdi), %rdi ffffffff818d3531: e8 1a ff ff ff callq 0xffffffff818d3450 <__sys_connect> ffffffff818d3536: 48 98 cltq ffffffff818d3538: c3 retq ffffffff818d3539: 0f 1f 80 00 00 00 00 nopl (%rax) As the assembly code for __x64_sys_connect shows, parameters should be fetched and set into rdi, rsi, rdx registers prior to calling __sys_connect. Because of this problem, this commit fixes the sys_connect event by first getting the value of the rdi register and then the value of the rdi, rsi, and rdx register through an offset based on that value. Fixes: 34745aed ("samples/bpf: fix kprobe attachment issue on x64") Signed-off-by:
-
Stanislav Fomichev authored
Simple test that enforces a single SOCK_DGRAM socket per cgroup. Signed-off-by:
Stanislav Fomichev <sdf@google.com> Signed-off-by:
-
Stanislav Fomichev authored
Support attaching to BPF_CGROUP_INET_SOCK_RELEASE and properly display attach type upon prog dump. Signed-off-by:
-
Stanislav Fomichev authored
Add auto-detection for the cgroup/sock_release programs. Signed-off-by:
-
Stanislav Fomichev authored
Sometimes it's handy to know when the socket gets freed. In particular, we'd like to try to use a smarter allocation of ports for bpf_bind and explore the possibility of limiting the number of SOCK_DGRAM sockets the process can have. Implement BPF_CGROUP_INET_SOCK_RELEASE hook that triggers on inet socket release. It triggers only for userspace sockets (not in-kernel ones) and therefore has the same semantics as the existing BPF_CGROUP_INET_SOCK_CREATE. Signed-off-by:
-
Matteo Croce authored
priv->page_pool is an array, so comparing against it will always return true. Do a meaningful check by checking priv->page_pool[0] instead. While at it, clear the page_pool pointers on deallocation, or when an allocation error happens during init. Reported-by:
Colin Ian King <colin.king@canonical.com> Fixes: c2d6fe61 ("mvpp2: XDP TX support") Signed-off-by:
Matteo Croce <mcroce@microsoft.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Florian Westphal authored
We can re-use the existing work queue to handle path management instead of a dedicated work queue. Just move pm_worker to protocol.c, call it from the mptcp worker and get rid of the msk lock (already held). Signed-off-by:
Florian Westphal <fw@strlen.de> Reviewed-by:
Mat Martineau <mathew.j.martineau@linux.intel.com> Signed-off-by:
-
Wei Yongjun authored
In certain configurations without power management support, gcc report the following warning: drivers/net/ethernet/sun/cassini.c:5206:12: warning: 'cas_resume' defined but not used [-Wunused-function] 5206 | static int cas_resume(struct device *dev_d) | ^~~~~~~~~~ Mark cas_resume() as __maybe_unused to make it clear. Fixes: f193f4eb ("sun/cassini: use generic power management") Reported-by:Hulk Robot <hulkci@huawei.com> Signed-off-by:
Wei Yongjun <weiyongjun1@huawei.com> Signed-off-by:
-
Vaibhav Gupta authored
The upgraded .suspend() and .resume() throw "defined but not used [-Wunused-function]" warning for certain configurations. Mark them with "__maybe_unused" attribute. Compile-tested only. Fixes: b0db0cc2 ("sun/niu: use generic power management") Reported-by:
Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by:
Vaibhav Gupta <vaibhavgupta40@gmail.com> Signed-off-by:
-
