- May 17, 2023
-
-
Andrii Nakryiko authored
It's trivial for user to trigger "verifier log line truncated" warning, as verifier has a fixed-sized buffer of 1024 bytes (as of now), and there are at least two pieces of user-provided information that can be output through this buffer, and both can be arbitrarily sized by user: - BTF names; - BTF.ext source code lines strings. Verifier log buffer should be properly sized for typical verifier state output. But it's sort-of expected that this buffer won't be long enough in some circumstances. So let's drop the check. In any case code will work correctly, at worst truncating a part of a single line output. Reported-by:
<syzbot+8b2a08dfbd25fd933d75@syzkaller.appspotmail.com> Signed-off-by:
Andrii Nakryiko <andrii@kernel.org> Link: https://lore.kernel.org/r/20230516180409.3549088-1-andrii@kernel.org Signed-off-by:
Alexei Starovoitov <ast@kernel.org>
-
Alexei Starovoitov authored
Magnus Karlsson says: ==================== Prepare the AF_XDP selftests test framework code for the upcoming multi-buffer support in AF_XDP. This so that the multi-buffer patch set does not become way too large. In that upcoming patch set, we are only including the multi-buffer tests together with any framework code that depends on the new options bit introduced in the AF_XDP multi-buffer implementation itself. Currently, the test framework is based on the premise that a packet consists of a single fragment and thus occupies a single buffer and a single descriptor. Multi-buffer breaks this assumption, as that is the whole purpose of it. Now, a packet can consist of multiple buffers and therefore consume multiple descriptors. The patch set starts with some clean-ups and simplifications followed by patches that make sure that the current code works even when a packet occupies multiple buffers. The actual code for sending and receiving multi-buffer packets will be included in the AF_XDP multi-buffer patch set as it depends on a new bit being used in the options field of the descriptor. Patch set anatomy: 1: The XDP program was unnecessarily changed many times. Fixes this. 2: There is no reason to generate a full UDP/IPv4 packet as it is never used. Simplify the code by just generating a valid Ethernet frame. 3: Introduce a more complicated payload pattern that can detect fragments out of bounds in a multi-buffer packet and other errors found in single-fragment packets. 4: As a convenience, dump the content of the faulty packet at error. 5: To simplify the code, make the usage of the packet stream for Tx and Rx more similar. 6: Store the offset of the packet in the buffer in the struct pkt definition instead of the address in the umem itself and introduce a simple buffer allocator. The address only made sense when all packets consumed a single buffer. Now, we do not know beforehand how many buffers a packet will consume, so we instead just allocate a buffer from the allocator and specify the offset within that buffer. 7: Test for huge pages only once instead of before each test that needs it. 8: Populate the fill ring based on how many frags are needed for each packet. 9: Change the data generation code so it can generate data for multi-buffer packets too. 10: Adjust the packet pacing algorithm so that it can cope with multi-buffer packets. The pacing algorithm is present so that Tx does not send too many packets/frames to Rx that it starts to drop packets. That would ruin the tests. v1 -> v2: * Fixed spelling error in patch #6 [Simon] * Fixed compilation error with llvm in patch #7 [Daniel] Thanks: Magnus ==================== Signed-off-by: -
Magnus Karlsson authored
Modify the packet pacing algorithm so that it works with multi-buffer packets. This algorithm makes sure we do not send too many buffers to the receiving thread so that packets have to be dropped. The previous algorithm made the assumption that each packet only consumes one buffer, but that is not true anymore when multi-buffer support gets added. Instead, we find out what the largest packet size is in the packet stream and assume that each packet will consume this many buffers. This is conservative and overly cautious as there might be smaller packets in the stream that need fewer buffers per packet. But it keeps the algorithm simple. Also simplify it by removing the pthread conditional and just test if there is enough space in the Rx thread before trying to send one more batch. Also makes the tests run faster. Signed-off-by:
Magnus Karlsson <magnus.karlsson@intel.com> Link: https://lore.kernel.org/r/20230516103109.3066-11-magnus.karlsson@gmail.com Signed-off-by:
-
Magnus Karlsson authored
Add the ability to generate data in the packets that are correct for multi-buffer packets. The ethernet header should only go into the first fragment followed by data and the others should only have data. We also need to modify the pkt_dump function so that it knows what fragment has an ethernet header so it can print this. Signed-off-by:
-
Magnus Karlsson authored
Populate the fill ring based on the number of frags a packet needs. With multi-buffer support, a packet might require more than a single fragment/buffer, so the function xsk_populate_fill_ring() needs to consider how many buffers a packet will consume, and put that many buffers on the fill ring for each packet it should receive. As we are still not sending any multi-buffer packets, the function will only produce one buffer per packet at the moment. Signed-off-by:
-
Magnus Karlsson authored
Test for hugepages only once at the beginning of the execution of the whole test suite, instead of before each test that needs huge pages. These are the tests that use unaligned mode. As more unaligned tests will be added, so the current system just does not scale. With this change, there are now three possible outcomes of a test run: fail, pass, or skip. To simplify the handling of this, the function testapp_validate_traffic() now returns this value to the main loop. As this function is used by nearly all tests, it meant a small change to most of them. Signed-off-by:
-
Magnus Karlsson authored
Store the offset in struct pkt instead of the address. This is important since address is only meaningful in the context of a packet that is stored in a single umem buffer and thus a single Tx descriptor. If the packet, in contrast need to be represented by multiple buffers in the umem, storing the address makes no sense since the packet will consist of multiple buffers in the umem at various addresses. This change is in preparation for the upcoming multi-buffer support in AF_XDP and the corresponding tests. So instead of indicating the address, we instead indicate the offset of the packet in the first buffer. The actual address of the buffer is allocated from the umem with a new function called umem_alloc_buffer(). This also means we can get rid of the use_fill_for_addr flag as the addresses fed into the fill ring will always be the offset from the pkt specification in the packet stream plus the address of the allocated buffer from the umem. No special casing needed. Signed-off-by:
-
Magnus Karlsson authored
Convert the current variable rx_pkt_nb to an iterator that can be used for both Rx and Tx. This to simplify the code and making Tx more like Rx that already has this feature. Signed-off-by:
-
Magnus Karlsson authored
Dump the content of the packet when a test finds that packets are received out of order, the length is wrong, or some other packet error. Use the already existing pkt_dump function for this and call it when the above errors are detected. Get rid of the command line option for dumping packets as it is not useful to print out thousands of good packets followed by the faulty one you would like to see. Signed-off-by:
-
Magnus Karlsson authored
Add a varying payload pattern within the packet. Instead of having just a packet number that is the same for all words in a packet, make each word different in the packet. The upper 16-bits are set to the packet number and the lower 16-bits are the sequence number of the words in this packet. So the 3rd packet's 5th 32-bit word of data will contain the number (2<<32) | 4 as they are numbered from 0. This will make it easier to detect fragments that are out of order when starting to test multi-buffer support. The member payload in the packet is renamed pkt_nb to reflect that it is now only a pkt_nb, not the real payload as seen above. Signed-off-by:
-
Magnus Karlsson authored
Implement support for generating pkts with variable length. Before this patch, they were all 64 bytes, exception for some packets of zero length and some that were too large. This feature will be used to test multi-buffer support for which large packets are needed. The packets are also made simpler, just a valid Ethernet header followed by a sequence number. This so that it will become easier to implement packet generation when each packet consists of multiple fragments. There is also a maintenance burden associated with carrying all this code for generating proper UDP/IP packets, especially since they are not needed. Signed-off-by:
-
Magnus Karlsson authored
Do not change the XDP program for the Tx thread when not needed. It was erroneously compared to the XDP program for the Rx thread, which is always going to be different, which meant that the code made unnecessary switches to the same program it had before. This did not affect functionality, just performance. Signed-off-by:
-
Alexei Starovoitov authored
Jiri Olsa says: ==================== hi, I noticed several times in discussions that we should move test kfuncs into kernel module, now perhaps even more pressing with all the kfunc effort. This patchset moves all the test kfuncs into bpf_testmod. I added bpf_testmod/bpf_testmod_kfunc.h header that is shared between bpf_testmod kernel module and BPF programs. v4 changes: - s390 supports long calls [1] now, so it can call now kfuncs from module [Ilya] - added acks [David] - cleanups for ptr_to_u64 function [David] - use relative path for bpf_testmod_kfunc.h include [Andrii] - new libbpf fix (patch 1) for gen_loader v3 changes: - added acks [David] - added bpf_testmod.ko make dependency for bpf test progs [David] - better handling of __ksym and refcount_t in bpf_testmod_kfunc.h [David] - removed 'extern' from kfuncs declarations [David] - typo in header guard macro [David] - use only stdout in un/load_bpf_testmod v2 changes: - add 74bc3a5a into bpf-next/master CI, so the test would pass https://github.com/kernel-patches/vmtest/pull/192 - remove extra externs [Artem] - using un/load_bpf_testmod in other tests - rebased thanks, jirka [1] 1cf3bfc6 bpf: Support 64-bit pointers to kfuncs ==================== Signed-off-by:
-
Jiri Olsa authored
Moving kernel test kfuncs into bpf_testmod kernel module, and adding necessary init calls and BTF IDs records. We need to keep following structs in kernel: struct prog_test_ref_kfunc struct prog_test_member (embedded in prog_test_ref_kfunc) The reason is because they need to be marked as rcu safe (check test prog mark_ref_as_untrusted_or_null) and such objects are being required to be defined only in kernel at the moment (see rcu_safe_kptr check in kernel). We need to keep also dtor functions for both objects in kernel: bpf_kfunc_call_test_release bpf_kfunc_call_memb_release We also keep the copy of these struct in bpf_testmod_kfunc.h, because other test functions use them. This is unfortunate, but this is just temporary solution until we are able to these structs them to bpf_testmod completely. As suggested by David adding bpf_testmod.ko make dependency for bpf programs, so they are rebuilt if we change the bpf_testmod.ko module. Also adding missing __bpf_kfunc to bpf_kfunc_call_test4 functions. Signed-off-by:
Jiri Olsa <jolsa@kernel.org> Acked-by:
David Vernet <void@manifault.com> Link: https://lore.kernel.org/r/20230515133756.1658301-11-jolsa@kernel.org Signed-off-by:
-
Jiri Olsa authored
There's no need to keep the extern in kfuncs declarations. Suggested-by:
-
Jiri Olsa authored
Currently the test_verifier allows test to specify kfunc symbol and search for it in the kernel BTF. Adding the possibility to search for kfunc also in bpf_testmod module when it's not found in kernel BTF. To find bpf_testmod btf we need to get back SYS_ADMIN cap. Acked-by:
-
Jiri Olsa authored
Loading bpf_testmod kernel module for verifier test. We will move all the tests kfuncs into bpf_testmod in following change. Acked-by:
-
Jiri Olsa authored
Now that we have un/load_bpf_testmod helpers in testing_helpers.h, we can use it in other tests and save some lines. Acked-by:
-
Jiri Olsa authored
Do not unload bpf_testmod in load_bpf_testmod, instead call unload_bpf_testmod separatelly. This way we will be able use un/load_bpf_testmod functions in other tests that un/load bpf_testmod module. Acked-by:
-
Jiri Olsa authored
We are about to use un/load_bpf_testmod functions in couple tests and it's better to print output to stdout, so it's aligned with tests ASSERT macros output, which use stdout as well. Acked-by:
-
Jiri Olsa authored
Moving test_progs helpers to testing_helpers object so they can be used from test_verifier in following changes. Also adding missing ifndef header guard to testing_helpers.h header. Using stderr instead of env.stderr because un/load_bpf_testmod helpers will be used outside test_progs. Also at the point of calling them in test_progs the std files are not hijacked yet and stderr is the same as env.stderr. Acked-by:
-
Jiri Olsa authored
Move all kfunc exports into separate bpf_testmod_kfunc.h header file and include it in tests that need it. We will move all test kfuncs into bpf_testmod in following change, so it's convenient to have declarations in single place. The bpf_testmod_kfunc.h is included by both bpf_testmod and bpf programs that use test kfuncs. As suggested by David, the bpf_testmod_kfunc.h includes vmlinux.h and bpf/bpf_helpers.h for bpf programs build, so the declarations have proper __ksym attribute and we can resolve all the structs. Note in kfunc_call_test_subprog.c we can no longer use the sk_state define from bpf_tcp_helpers.h (because it clashed with vmlinux.h) and we need to address __sk_common.skc_state field directly. Acked-by:
-
Jiri Olsa authored
When moving some of the test kfuncs to bpf_testmod I hit an issue when some of the kfuncs that object uses are in module and some in vmlinux. The problem is that both vmlinux and module kfuncs get allocated btf_fd_idx index into fd_array, but we store to it the BTF fd value only for module's kfunc, not vmlinux's one because (it's zero). Then after the program is loaded we check if fd_array[btf_fd_idx] != 0 and close the fd. When the object has kfuncs from both vmlinux and module, the fd from fd_array[btf_fd_idx] from previous load will be stored in there for vmlinux's kfunc, so we close unrelated fd (of the program we just loaded in my case). Fixing this by storing zero to fd_array[btf_fd_idx] for vmlinux kfuncs, so the we won't close stale fd. Signed-off-by:
-
Yonghong Song authored
llvm patch [1] enabled cross-function optimization for func arguments (ArgumentPromotion) at -O2 level. And this caused s390 sock_fields test failure ([2]). The failure is gone right now as patch [1] was reverted in [3]. But it is possible that patch [3] will be reverted again and then the test failure in [2] will show up again. So it is desirable to fix the failure regardless. The following is an analysis why sock_field test fails with llvm patch [1]. The main problem is in static __noinline bool sk_dst_port__load_word(struct bpf_sock *sk) { __u32 *word = (__u32 *)&sk->dst_port; return word[0] == bpf_htons(0xcafe); } static __noinline bool sk_dst_port__load_half(struct bpf_sock *sk) { __u16 *half = (__u16 *)&sk->dst_port; return half[0] == bpf_htons(0xcafe); } ... int read_sk_dst_port(struct __sk_buff *skb) { ... sk = skb->sk; ... if (!sk_dst_port__load_word(sk)) RET_LOG(); if (!sk_dst_port__load_half(sk)) RET_LOG(); ... } Through some cross-function optimization by ArgumentPromotion optimization, the compiler does: static __noinline bool sk_dst_port__load_word(__u32 word_val) { return word_val == bpf_htons(0xcafe); } static __noinline bool sk_dst_port__load_half(__u16 half_val) { return half_val == bpf_htons(0xcafe); } ... int read_sk_dst_port(struct __sk_buff *skb) { ... sk = skb->sk; ... __u32 *word = (__u32 *)&sk->dst_port; __u32 word_val = word[0]; ... if (!sk_dst_port__load_word(word_val)) RET_LOG(); __u16 half_val = word_val >> 16; if (!sk_dst_port__load_half(half_val)) RET_LOG(); ... } In current uapi bpf.h, we have struct bpf_sock { ... __be16 dst_port; /* network byte order */ __u16 :16; /* zero padding */ ... }; But the old kernel (e.g., 5.6) we have struct bpf_sock { ... __u32 dst_port; /* network byte order */ ... }; So for backward compatability reason, 4-byte load of dst_port is converted to 2-byte load internally. Specifically, 'word_val = word[0]' is replaced by 2-byte load by the verifier and this caused the trouble for later sk_dst_port__load_half() where half_val becomes 0. Typical usr program won't have such a code pattern tiggering the above bug, so let us fix the test failure with source code change. Adding an empty asm volatile statement seems enough to prevent undesired transformation. [1] https://reviews.llvm.org/D148269 [2] https://lore.kernel.org/bpf/e7f2c5e8-a50c-198d-8f95-388165f1e4fd@meta.com/ [3] https://reviews.llvm.org/rG141be5c062ecf22bd287afffd310e8ac4711444a Tested-by:Ilya Leoshkevich <iii@linux.ibm.com> Signed-off-by:
Yonghong Song <yhs@fb.com> Link: https://lore.kernel.org/r/20230516214945.1013578-1-yhs@fb.com Signed-off-by:
-
Andrii Nakryiko authored
Change netcnt to demand at least 10K packets, as we frequently see some stray packet arriving during the test in BPF CI. It seems more important to make sure we haven't lost any packet than enforcing exact number of packets. Cc: Stanislav Fomichev <sdf@google.com> Signed-off-by:
Stanislav Fomichev <sdf@google.com> Link: https://lore.kernel.org/r/20230515204833.2832000-1-andrii@kernel.org Signed-off-by:
-
Jakub Kicinski authored
This reverts commit b2cbac9b . We have multiple reports of obvious breakage from this patch. Reported-by:
Ido Schimmel <idosch@idosch.org> Link: https://lore.kernel.org/all/ZGIRWjNcfqI8yY8W@shredder/ Link: https://lore.kernel.org/all/CADJHv_sDK=0RrMA2FTZQV5fw7UQ+qY=HG21Wu5qb0V9vvx5w6A@mail.gmail.com/ Reported-by:
<syzbot+a5e719ac7c268e414c95@syzkaller.appspotmail.com> Reported-by:
<syzbot+a03fd670838d927d9cd8@syzkaller.appspotmail.com> Fixes: b2cbac9b ("net: Remove low_thresh in ip defrag") Link: https://lore.kernel.org/r/20230517034112.1261835-1-kuba@kernel.org Signed-off-by:
Jakub Kicinski <kuba@kernel.org>
-
https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-nextJakub Kicinski authored
Daniel Borkmann says: ==================== pull-request: bpf-next 2023-05-16 We've added 57 non-merge commits during the last 19 day(s) which contain a total of 63 files changed, 3293 insertions(+), 690 deletions(-). The main changes are: 1) Add precision propagation to verifier for subprogs and callbacks, from Andrii Nakryiko. 2) Improve BPF's {g,s}setsockopt() handling with wrong option lengths, from Stanislav Fomichev. 3) Utilize pahole v1.25 for the kernel's BTF generation to filter out inconsistent function prototypes, from Alan Maguire. 4) Various dyn-pointer verifier improvements to relax restrictions, from Daniel Rosenberg. 5) Add a new bpf_task_under_cgroup() kfunc for designated task, from Feng Zhou. 6) Unblock tests for arm64 BPF CI after ftrace supporting direct call, from Florent Revest. 7) Add XDP hint kfunc metadata for RX hash/timestamp for igc, from Jesper Dangaard Brouer....
-
- May 16, 2023
-
-
Paolo Abeni authored
Bagas Sanjaya says: ==================== SPDX conversion for bonding, 8390, and i825xx drivers This series is SPDX conversion for bonding, 8390, and i825xx driver subsystems. It is splitted from v2 of my SPDX conversion series in response to Didi's GPL full name fixes [1] to make it easily digestible. The conversion in this series is divided by each subsystem and by license type. [1]: https://lore.kernel.org/linux-spdx/20230512100620.36807-1-bagasdotme@gmail.com/ ==================== Link: https://lore.kernel.org/r/20230515060714.621952-1-bagasdotme@gmail.com Signed-off-by:
Paolo Abeni <pabeni@redhat.com>
-
Bagas Sanjaya authored
The boilerplate reads that sun3_8256 driver is an extension to Linux kernel core, hence add SPDX license identifier for GPL 2.0. Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Michael Hipp <hippm@informatik.uni-tuebingen.de> Cc: Sam Creasey <sammy@sammy.net> Signed-off-by:
Bagas Sanjaya <bagasdotme@gmail.com> Signed-off-by:
-
Bagas Sanjaya authored
Replace unversioned GPL boilerplate notice with corresponding SPDX license identifier, which is GPL 1.0+. Cc: Donald Becker <becker@scyld.com> Cc: Richard Hirst <richard@sleepie.demon.co.uk> Cc: Sam Creasey <sammy@sammy.net> Signed-off-by:
-
Bagas Sanjaya authored
The boilerplate refers to COPYING in the top-level directory of kernel tree. Replace it with corresponding SPDX license identifier. Cc: Donald Becker <becker@scyld.com> Cc: Peter De Schrijver <p2@mind.be> Cc: Topi Kanerva <topi@susanna.oulu.fi> Cc: Alain Malek <Alain.Malek@cryogen.com> Cc: Bruce Abbott <bhabbott@inhb.co.nz> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Richard Fontana <rfontana@redhat.com> Acked-by:
Greg Ungerer <gerg@linux-m68k.org> Signed-off-by:
-
Bagas Sanjaya authored
Replace boilerplate notice for unversioned GPL to SPDX tag for GPL 1.0+. For ne2k-pci.c, only add SPDX tag and keep the boilerplate instead, since the boilerplate notes that it must be preserved. Cc: David A. Hinds <dahinds@users.sourceforge.net> Cc: Donald Becker <becker@scyld.com> Cc: Alan Cox <alan@linux.intel.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Richard Fontana <rfontana@redhat.com> Signed-off-by:
-
Bagas Sanjaya authored
Previous batches of SPDX conversion missed bond_main.c and bonding_priv.h because these files doesn't mention intended GPL version. Add SPDX identifier to these files, assuming GPL 1.0+. Cc: Thomas Davis <tadavis@lbl.gov> Cc: Christophe JAILLET <christophe.jaillet@wanadoo.fr> Cc: Stephen Hemminger <stephen@networkplumber.org> Reviewed-by:
Simon Horman <simon.horman@corigine.com> Signed-off-by:
-
Yunsheng Lin authored
__skb_fill_page_desc_noacc() is not doing any pfmemalloc propagating, and yet it has a comment about that, commit 84ce071e ("net: introduce __skb_fill_page_desc_noacc") may have accidentally moved it to __skb_fill_page_desc_noacc(), so move it back to __skb_fill_page_desc() which is supposed to be doing pfmemalloc propagating. Signed-off-by:
Yunsheng Lin <linyunsheng@huawei.com> CC: Pavel Begunkov <asml.silence@gmail.com> Reviewed-by:
Pavel Begunkov <asml.silence@gmail.com> Link: https://lore.kernel.org/r/20230515050107.46397-1-linyunsheng@huawei.com Signed-off-by:
-
Yafang Shao authored
If it fails to attach fentry, the allocated bpf trampoline image will be left in the system. That can be verified by checking /proc/kallsyms. This meamleak can be verified by a simple bpf program as follows: SEC("fentry/trap_init") int fentry_run() { return 0; } It will fail to attach trap_init because this function is freed after kernel init, and then we can find the trampoline image is left in the system by checking /proc/kallsyms. $ tail /proc/kallsyms ffffffffc0613000 t bpf_trampoline_6442453466_1 [bpf] ffffffffc06c3000 t bpf_trampoline_6442453466_1 [bpf] $ bpftool btf dump file /sys/kernel/btf/vmlinux | grep "FUNC 'trap_init'" [2522] FUNC 'trap_init' type_id=119 linkage=static $ echo $((6442453466 & 0x7fffffff)) 2522 Note that there are two left bpf trampoline images, that is because the libbpf will fallback to raw tracepoint if -EINVAL is returned. Fixes: e21aa341 ("bpf: Fix fexit trampoline.") Signed-off-by:Yafang Shao <laoar.shao@gmail.com> Signed-off-by:
Daniel Borkmann <daniel@iogearbox.net> Acked-by:
Song Liu <song@kernel.org> Cc: Jiri Olsa <olsajiri@gmail.com> Link: https://lore.kernel.org/bpf/20230515130849.57502-2-laoar.shao@gmail.com
-
Yafang Shao authored
After commit e21aa341 ("bpf: Fix fexit trampoline."), the selector is only used to indicate how many times the bpf trampoline image are updated and been displayed in the trampoline ksym name. After the trampoline is freed, the selector will start from 0 again. So the selector is a useless value to the user. We can remove it. If the user want to check whether the bpf trampoline image has been updated or not, the user can compare the address. Each time the trampoline image is updated, the address will change consequently. Jiri also pointed out another issue that perf is still using the old name "bpf_trampoline_%lu", so this change can fix the issue in perf. Fixes: e21aa341 ("bpf: Fix fexit trampoline.") Signed-off-by:
-
Florent Revest authored
This extends the BPF trampoline JIT to support attachment to functions that take small structures (up to 128bit) as argument. This is trivially achieved by saving/restoring a number of "argument registers" rather than a number of arguments. The AAPCS64 section 6.8.2 describes the parameter passing ABI. "Composite types" (like C structs) below 16 bytes (as enforced by the BPF verifier) are provided as part of the 8 argument registers as explained in the section C.12. Signed-off-by:
Florent Revest <revest@chromium.org> Signed-off-by:
Xu Kuohai <xukuohai@huawei.com> Link: https://lore.kernel.org/bpf/20230511140507.514888-1-revest@chromium.org
-
Alan Maguire authored
On aarch64, "bpftool feature" reports an incorrect BPF JIT limit: $ sudo /sbin/bpftool feature Scanning system configuration... bpf() syscall restricted to privileged users JIT compiler is enabled JIT compiler hardening is disabled JIT compiler kallsyms exports are enabled for root skipping kernel config, can't open file: No such file or directory Global memory limit for JIT compiler for unprivileged users is -201326592 bytes This is because /proc/sys/net/core/bpf_jit_limit reports $ sudo cat /proc/sys/net/core/bpf_jit_limit 68169519595520 ...and an int is assumed in read_procfs(). Change read_procfs() to return a long to avoid negative value reporting. Fixes: 7a4522bb ("tools: bpftool: add probes for /proc/ eBPF parameters") Reported-by:
Nicky Veitch <nicky.veitch@oracle.com> Signed-off-by:
Alan Maguire <alan.maguire@oracle.com> Signed-off-by:
Quentin Monnet <quentin@isovalent.com> Link: https://lore.kernel.org/bpf/20230512113134.58996-1-alan.maguire@oracle.com
-
Andrii Nakryiko authored
Subsequent instruction index (subseq_idx) is an index of an instruction that was verified/executed by verifier after the currently processed instruction. It is maintained during precision backtracking processing and is used to detect various subprog calling conditions. This patch fixes the bug with incorrectly resetting subseq_idx to -1 when going from child state to parent state during backtracking. If we don't maintain correct subseq_idx we can misidentify subprog calls leading to precision tracking bugs. One such case was triggered by test_global_funcs/global_func9 test where global subprog call happened to be the very last instruction in parent state, leading to subseq_idx==-1, triggering WARN_ONCE: [ 36.045754] verifier backtracking bug [ 36.045764] WARNING: CPU: 13 PID: 2073 at kernel/bpf/verifier.c:3503 __mark_chain_precision+0xcc6/0xde0 [ 36.046819] Modules linked in: aesni_intel(E) crypto_simd(E) cryptd(E) kvm_intel(E) kvm(E) irqbypass(E) i2c_piix4(E) serio_raw(E) i2c_core(E) crc32c_intel) [ 36.048040] CPU: 13 PID: 2073 Comm: test_progs Tainted: G W OE 6.3.0-07976-g4d585f48 -dirty #972 [ 36.048783] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.12.0-59-gc9ba5276e321-prebuilt.qemu.org 04/01/2014 [ 36.049648] RIP: 0010:__mark_chain_precision+0xcc6/0xde0 [ 36.050038] Code: 3d 82 c6 05 bb 35 32 02 01 e8 66 21 ec ff 0f 0b b8 f2 ff ff ff e9 30 f5 ff ff 48 c7 c7 f3 61 3d 82 4c 89 0c 24 e8 4a 21 ec ff <0f> 0b 4c0 With the fix precision tracking across multiple states works correctly now: mark_precise: frame0: last_idx 45 first_idx 38 subseq_idx -1 mark_precise: frame0: regs=r8 stack= before 44: (61) r7 = *(u32 *)(r10 -4) mark_precise: frame0: regs=r8 stack= before 43: (85) call pc+41 mark_precise: frame0: regs=r8 stack= before 42: (07) r1 += -48 mark_precise: frame0: regs=r8 stack= before 41: (bf) r1 = r10 mark_precise: frame0: regs=r8 stack= before 40: (63) *(u32 *)(r10 -48) = r1 mark_precise: frame0: regs=r8 stack= before 39: (b4) w1 = 0 mark_precise: frame0: regs=r8 stack= before 38: (85) call pc+38 mark_precise: frame0: parent state regs=r8 stack=: R0_w=scalar() R1_w=map_value(off=4,ks=4,vs=8,imm=0) R6=1 R7_w=scalar() R8_r=P0 R10=fpm mark_precise: frame0: last_idx 36 first_idx 28 subseq_idx 38 mark_precise: frame0: regs=r8 stack= before 36: (18) r1 = 0xffff888104f2ed14 mark_precise: frame0: regs=r8 stack= before 35: (85) call pc+33 mark_precise: frame0: regs=r8 stack= before 33: (18) r1 = 0xffff888104f2ed10 mark_precise: frame0: regs=r8 stack= before 32: (85) call pc+36 mark_precise: frame0: regs=r8 stack= before 31: (07) r1 += -4 mark_precise: frame0: regs=r8 stack= before 30: (bf) r1 = r10 mark_precise: frame0: regs=r8 stack= before 29: (63) *(u32 *)(r10 -4) = r7 mark_precise: frame0: regs=r8 stack= before 28: (4c) w7 |= w0 mark_precise: frame0: parent state regs=r8 stack=: R0_rw=scalar() R6=1 R7_rw=scalar() R8_rw=P0 R10=fp0 fp-48_r=mmmmmmmm mark_precise: frame0: last_idx 27 first_idx 16 subseq_idx 28 mark_precise: frame0: regs=r8 stack= before 27: (85) call pc+31 mark_precise: frame0: regs=r8 stack= before 26: (b7) r1 = 0 mark_precise: frame0: regs=r8 stack= before 25: (b7) r8 = 0 Note how subseq_idx starts out as -1, then is preserved as 38 and then 28 as we go up the parent state chain. Reported-by:
-
- May 15, 2023
-
-
Dave Marchevsky authored
For kfuncs like bpf_obj_drop and bpf_refcount_acquire - which take user-defined types as input - the verifier needs to track the specific type passed in when checking a particular kfunc call. This requires tracking (btf, btf_id) tuple. In commit 7c50b1cb ("bpf: Add bpf_refcount_acquire kfunc") I added an anonymous union with inner structs named after the specific kfuncs tracking this information, with the goal of making it more obvious which kfunc this data was being tracked / expected to be tracked on behalf of. In a recent series adding a new user of this tuple, Alexei mentioned that he didn't like this union usage as it doesn't really help with readability or bug-proofing ([0]). In an offline convo we agreed to have the tuple be fields (arg_btf, arg_btf_id), with comments in bpf_kfunc_call_arg_meta definition enumerating the uses of the fields by kfunc-specific handling logic. Such a pattern is used by struct bpf_reg_state without trouble. Accordingly, this patch removes the anonymous union in favor of arg_btf and arg_btf_id fields and comment enumerating their current uses. The patch also removes struct btf_and_id, which was only being used by the removed union's inner structs. This is a mechanical change, existing linked_list and rbtree tests will validate that correct (btf, btf_id) are being passed. [0]: https://lore.kernel.org/bpf/20230505021707.vlyiwy57vwxglbka@dhcp-172-26-102-232.dhcp.thefacebook.com Signed-off-by:
Dave Marchevsky <davemarchevsky@fb.com> Link: https://lore.kernel.org/r/20230510213047.1633612-1-davemarchevsky@fb.com Signed-off-by:
-
