aboutsummaryrefslogtreecommitdiffstats
path: root/executor
Commit message (Collapse)AuthorAgeFilesLines
* executor/kvm: bug fix and minor refactor in KVMMarios Pomonis2025-03-274-328/+326
| | | | | | | | | | | | | | | | * Fixes a bug when setting up a 64-bit guest by making the bit manipulation macros produce unsigned long long: To create a VCPU that has paging enabled, one needs to set the CR0.PE and CR0.PG bits in CR0. The latter is problematic when setting up a 64-bit guest since if the macro is not using 1ULL, it sign extends the output (in 64-bit mode the control registers are extended to 64-bits with some of the CR0[32:63] bits reserved). This results in either failing the KVM_SET_SREGS ioctl (in newer kernel versions) or just failing the KVM_RUN ioctl with EXIT_REASON_INVALID_STATE. * Moved the bit manipulation definitions from the amd64 specific to the generic kvm header to consolidate them with the already existing ones. Prefixed them with X86_ to avoid confusion.
* executor: pull in errno.h on BSD systemsMark Johnston2025-02-261-0/+1
| | | | | The error handling for the setsid() call in sandbox_common() requires it. Without it, some csource builds fail.
* executor: fix cover_protect() on FreeBSDMark Johnston2025-02-212-2/+8
| | | | | | | | | | | | | | | | | During machine checks, syzkaller will execute calls with coverage disabled, in which case per-thread coverage structures are zeroed out. write_output() will temporarily map the coverage data as writeable via CoverAccessScope, whether or not cover is enabled. In effect, write_output() may trigger a call mprotect(0, kCoverSize, PROT_RW). On FreeBSD, mprotect() silently ignores unmapped regions, so this does not result in an error. In fact, kCoverSize is now large enough that this ends up removing the eXecute bit from part of syz-executor's text region. Make CoverAccessScope a no-op if coverage is not enabled. Modify BSD cover_protect() and cover_unprotect() to fail if invoked when coverage is disabled.
* pkg/rpcserver: pkg/flatrpc: executor: add handshake stage 0Alexander Potapenko2025-02-201-0/+16
| | | | | | | | | | | | | | | | | | | | As we figured out in #5805, syz-manager treats random incoming RPC connections as trusted, and will crash if a non-executor client sends an invalid packet to it. To address this issue, we introduce another stage of handshake, which includes a cookie exchange: - upon connection from an executor, the manager sends a ConnectHello RPC message to it, which contains a random 64-bit cookie; - the executor calculates a hash of that cookie and includes it into its ConnectRequest together with the other information; - before checking the validity of ConnectRequest, the manager ensures client sanity (passed ID didn't change, hashed cookie has the expected value) We deliberately pick a random cookie instead of a magic number: if the fuzzer somehow learns to send packets to the manager, we don't want it to crash multiple managers on the same machine.
* executor: fix MAP_FIXED_NOREPLACE dependencyTaras Madan2025-02-111-0/+3
| | | | Some environments don't define MAP_FIXED_NOREPLACE.
* executor: favor MAP_FIXED_NOREPLACE over MAP_FIXEDAleksandr Nogikh2025-02-044-5/+13
| | | | | | | | | | | | MAP_FIXED_NOREPLACE allows to fail early if we happened to overlap with an existing memory mapping. It should help detects bugs #5674 at an earlier stage, before it led to memory corruptions. MAP_FIXED_NOREPLACE is supported from Linux 4.17, which is okay for all syzkaller use cases on syzbot. There's no such option for some of the supported OSes, so set it depending on the configuration we're building for.
* executor: increase timeouts for glob requestsNick Biryulin2025-01-281-1/+5
| | | | | | Proper glob resolution is required for fuzzing. If it times out, it does so silently, and fuzzing dictionary will be smaller then expected, without any obvious errors. Given that, it makes sense to increase glob timeouts.
* executor: mount /syz-inputs dir to fuzzing sandboxNick Biryulin2025-01-281-0/+10
| | | | | | | | Syzkaller allows user to specify filepath arguments in syscalls via globs. However, on linux, you are effectivly limited to some /sys and /dev paths due to sandboxing. With this change, user can supply their custom fuzzing artifacts to /syz-inputs to use those in globs. They are mounted read-only to increase reproducibility.
* all: remove more mentions of the vendor folderAleksandr Nogikh2025-01-231-2/+4
|
* executor: arm64: add SYZOS_API_MRSMarios Pomonis2025-01-141-0/+36
| | | | Add support for the MRS instruction in a similar manner to MSR.
* executor: arm64: fix compiler warnings on inline assemblyAlexander Potapenko2025-01-091-5/+5
| | | | | | Make sure operands passed to 64-bit MOV, MSR and MRS instructions are actually 64-bit. This fixes compiler warnings in certain build configurations.
* executor: pkg/vminfo: sys/linux: arm64: implement syz_kvm_assert_reg()Alexander Potapenko2024-12-182-1/+20
| | | | Add a pseudo-syscall to assert on register values.
* all: add support for binaries shipped with targetStefan Wiehler2024-12-171-3/+3
| | | | | | | | | | | In some build environments (notably Yocto), syzkaller host and target binaries end up in separate packages for each built architecture, which are then shipped with the respective image/SDK. Add the "Execprog/ExecutorBinOnTarget" and "StraceBinOnTarget" options to the manager config, which when set expects the respective binaries to be shipped with the target image and does not attempt to copy them from the host.
* executor: mount gadgetfsDmitry Vyukov2024-12-111-0/+17
| | | | We can reach it at least with automatic descriptions.
* executor: query globs in the test program contextDmitry Vyukov2024-12-115-37/+89
| | | | | | | | | | | | | | | | | We query globs for 2 reasons: 1. Expand glob types in syscall descriptions. 2. Dynamic file probing for automatic descriptions generation. In both of these contexts are are interested in files that will be present during test program execution (rather than normal unsandboxed execution). For example, some files may not be accessible to test programs after pivot root. On the other hand, we create and link some additional files for the test program that don't normally exist. Add a new request type for querying of globs that are executed in the test program context.
* pkg/vminfo: sys/linux: executor: define syz_kvm_assert_syzos_uexit()Alexander Potapenko2024-12-112-1/+20
| | | | | The new pseudo-syscall will serve as a test assertion, checking the uexit return value. This is going to help us validate SyzOS code.
* executor: arm: check for zero VM handle in syz_kvm_add_vcpu()Alexander Potapenko2024-12-061-1/+7
| | | | | | | | | When running syscalls asynchronously, syz_kvm_add_vcpu() sometimes receives a zero VM handle, on which it then crashes. Check for the zero value to ensure stability of the tests in sys/linux/tests. Also make sure to set errno for the pseudo-syscall in the cases where it's not done by the underlying syscalls.
* executor: setup sysctl/cgroups in snapshot modeDmitry Vyukov2024-12-061-0/+8
| | | | We missed that step for snapshot mode.
* executor: add test_syzos() for arm64Alexander Potapenko2024-12-052-2/+21
| | | | | Make sure regressions in guest code validation are reported during testing rather than fuzzing.
* executor: arm: disable test_globAlexander Potapenko2024-12-051-0/+6
| | | | | | | | Glob() doesn't work on 32-bit ARM when run on a 64-bit system under QEMU: https://gitlab.com/qemu-project/qemu/-/issues/263 Not sure whether this is specific to tests running under qemu-user, or the ARM32 executor in the wild as well.
* executor: arm64: rewrite guest_handle_its_send_cmd() without a switchAlexander Potapenko2024-12-051-31/+33
| | | | | | Prevent the compiler from generating a jump table by replacing a switch with a series of if statements. This is ugly, but lets us work around crashes caused by https://github.com/google/syzkaller/issues/5565
* executor: arm64: declare 'noinline' for SyzOSAlexander Potapenko2024-12-051-32/+35
| | | | | | Apply __attribute__((noinline)) to SyzOS API command handlers to prevent overly optimizing them. While at it, rearrange specifiers in guest function declarations
* executor: arm64: detect data relocations in SyzOSAlexander Potapenko2024-12-051-1/+28
| | | | | | | Detect and report ADRP instructions in the linked binaries to avoid crashes inside SyzOS. See https://github.com/google/syzkaller/issues/5565 for more context.
* executor: remove a tempnam() call from test.hAleksandr Nogikh2024-12-021-2/+0
| | | | It's no longer needed.
* sys/linux: updated fuse fs specificationsSablin Viacheslav2024-11-291-0/+7
|
* sys/linux: move some ARM-specific descriptions to a separate fileAlexander Potapenko2024-11-261-18/+0
| | | | | | | | | | | | | | This is done to solve a particular test failure running: $ tools/syz-env go test ./prog -run TestSpecialStructs , which failed on PPC64, because prog/rand.go instanciated a call to syz_kvm_setup_syzos_vm(), which requested too much memory (1024 pages) from the allocator (PPC64 uses 64k pages, so the number of available pages is lower). On the other hand, factoring out syzos-related descriptions is probably a nice thing to do anyway.
* executor: arm64: sys/linux: allocate 1024 pages for guest address spaceAlexander Potapenko2024-11-262-21/+12
| | | | | | | Pass 1024 pages of memory to both syz_kvm_setup_syzos_vm() and syz_kvm_setup_cpu$arm64() to make sure that: - there is enough memory for guest allocations (e.g. ITS pages) - host can tamper with that memory, provoking more bugs
* executor: arm64: implement SYZOS_API_ITS_SEND_CMDAlexander Potapenko2024-11-261-0/+113
| | | | | In addition to the predefined ITS setup, let the guest execute different ITS configuration commands in an attempt to trigger interesting interactions.
* executor: arm64: implement SYZOS_API_ITS_SETUPAlexander Potapenko2024-11-262-4/+444
| | | | | The new API call implements basic setup of the ARM Interrupt Translation Service for the given number of CPUs, virtual devices, and LPIs.
* executor: arm64: handle interrupt IDs above 1023Alexander Potapenko2024-11-261-5/+11
| | | | | | There's no need to mask the IDs, and it actually doesn't work for LPIs. Also add more comments.
* executor: arm64: allocate memory for ITS tables on the host sideAlexander Potapenko2024-11-263-1/+9
|
* executor: arm64: use KVM_PAGE_SIZE instead of 0x1000Alexander Potapenko2024-11-261-1/+1
|
* executor: add Glob testDmitry Vyukov2024-11-263-1/+92
|
* executor: don't revert coverage orderDmitry Vyukov2024-11-261-1/+2
| | | | | | | Currently we write coverage backwards. This is visible e.g. when running syz-execprog -coverfile, and in the manager raw cover mode. Write it in the right order.
* executor: improve globDmitry Vyukov2024-11-261-1/+25
| | | | | | Don't follow symlinks when globbing. It's haarmful for both files and dirs (see the added comment for details).
* executor: increase coverage buffer sizeDmitry Vyukov2024-11-203-9/+15
| | | | | | | | | | | The coverage buffer frequently overflows. We cannot increase it radically b/c they consume lots of memory (num procs x num kcovs x buffer size) and lead to OOM kills (at least with 8 procs and 2GB KASAN VM). So increase it 2x and slightly reduce number of threads/kcov descriptors. However, in snapshot mode we can be more aggressive (only 1 proc). This reduces number of overflows by ~~2-4x depending on syscall.
* pkg/manager: show number of times coverage for each call has overflowedDmitry Vyukov2024-11-202-2/+15
| | | | | If the overflows happen often, it's bad. Add visibility into this.
* executor: use any executor if the avoid mask included all of themAndrei Vagin2024-11-181-1/+13
| | | | | | | | | | | | | | | | After 9fc8fe026baa ("executor: better handling for hanged test processes"), yz-executor's responses may reference procids outside of the [0;procs] range. If procids are no longer dense on the syz-executor side, we cannot rely on this check in pkg/rpcserver: ``` if avoid == (uint64(1)<<runner.procs)-1 { avoid = 0 } ``` Signed-off-by: Andrei Vagin <avagin@google.com>
* executor: better handling for hanged test processesDmitry Vyukov2024-10-243-16/+75
| | | | | | | | | | | | | | | Currently we kill hanged processes and consider the corresponding test finished. We don't kill/wait for the actual test subprocess (we don't know its pid to kill, and waiting will presumably hang). This has 2 problems: 1. If the hanged process causes "task hung" report, we can't reproduce it, since the test finished too long ago (manager thinks its finished and discards the request). 2. The test process still consumed per-pid resources. Explicitly detect and handle such cases: Manager keeps these hanged tests forever, and we assign a new proc id for future processes (don't reuse the hanged one).
* executor: check ppid after setting PR_SET_PDEATHSIG for sandbox processDmitry Vyukov2024-10-241-0/+4
| | | | This helps to avoid leaking processes when killing races with PR_SET_PDEATHSIG.
* executor: don't create new session for sandbox processDmitry Vyukov2024-10-241-1/+0
| | | | | | | | It's unclear why we need a new session. Sessions group process groups, but we don't use that. Setsid also creates a new process group, but we don't kill this process group, so also unclear why this is needed.
* executor: don't kill executor exec process groupDmitry Vyukov2024-10-241-1/+0
| | | | | | | | | Killing a process group (negative pid) only makes sense when the process is a group leader (called setsid/setpgrp/setpgid). Executor exec process is not a group leader, so don't try to kill its group. For our controlled executor subprocesses we rely on PR_SET_PDEATHSIG for reliable killing of all child subprocesses.
* executor: sys/linux: arm64: reserve address for ITS, add a seed for ITS creationAlexander Potapenko2024-10-211-1/+4
| | | | | | | Reserve SYZOS address for the ITS redistributor at 0x08080000, add it to the list of kvm_guest_addrs. Also implement a syzlang test for the host part of ITS configuration as per https://www.kernel.org/doc/html/v6.1/virt/kvm/devices/arm-vgic-its.html
* executor: fix mounting of debugfs/smackfs/fusectl/binfmt_miscDmitry Vyukov2024-09-271-9/+14
| | | | | | | | All these broke when we started mounting new tmpfs for sandbox=root. Some are not mounted at all, some are mounted in the outer root and are not accessible from the new root. Mount then inside of the new root tmpfs. Other file systems (binderfs, cgroups) seem to be ok.
* executor: arm64: rename SYZ_KVM_* to KVM_*Alexander Potapenko2024-09-251-16/+16
| | | | | It is more common for the constants in the executor to not have the SYZ_ prefix.
* executor: arm64: store CPU ID in TPIDR_EL1Alexander Potapenko2024-09-252-4/+20
| | | | | | | | | Let SYZOS distinguish CPUs inside VM by storing their ID in TPIDR_EL1. Make sure existing code uses that ID: - in guest_handle_msr(), to ensure concurrent calls do not write to the same cache line; - in gicv3_irq_enable(), to ensure proper CPU ID is being used for IRQ setup.
* executor: arm64: sys/linux: implement syz_kvm_setup_syzos_vm and ↵Alexander Potapenko2024-09-253-54/+135
| | | | | | | | | | | | | | | | | | | | | | | syz_kvm_add_vcpu The old syz_kvm_setup_cpu() API mixed together VM and VCPU setup, making it harder to create and fuzz two VCPUs in the same VM. Introduce two new pseudo-syscalls, syz_kvm_setup_syzos_vm() and syz_kvm_add_vcpu(), that will simplify this task. syz_kvm_setup_syzos_vm() takes a VM file descriptor, performs VM setup (allocates guest memory and installs SYZOS code into it) and returns a new kvm_syz_vm resource, which is in fact a pointer to `struct kvm_syz_vm` encapsulating VM-specific data in the C code. syz_kvm_add_vcpu() takes the VM ID denoted by kvm_syz_vm and creates a new VCPU within that VM with a proper CPU number. It then stores the fuzzer-supplied SYZOS API sequence into the corresponding part (indexed by CPU number) of the VM memory slot, and sets up the CPU registers to interpret that sequence. The new pseudo-syscall let the fuzzer create independent CPUs that run different code sequences without interfering with each other.
* executor: arm64: accept cpu id in guest_main()Alexander Potapenko2024-09-251-2/+2
| | | | | Use the cpu id to choose the SYZOS API commands to be executed by this particular CPU.
* executor: arm64: factor out install_user_code()Alexander Potapenko2024-09-251-22/+42
| | | | | | | | Prepare to let multiple (up to 4) CPUs run different pieces of code by allocating 4 pages for ARM64_ADDR_USER_CODE. Pass the CPU id to guest_main(), so that it can pick the correct starting addres. syz_kvm_setup_cpu() will implicitly use cpuid=0 to retain its current functionality.
* executor: arm64: factor out setup_vm()Alexander Potapenko2024-09-251-34/+41
| | | | | | | | | No functional change. For multiple CPUs within the same VM, calls to syz_kvm_setup_cpu() will set up the VM memory space multiple times, so only the last one will take effect. Prepare to decouple VM setup from CPU setup by factoring this code out of syz_kvm_setup_cpu().
generated by ggit (джигит) mrf-deployment (git 2.46.0) at 2026-03-12 01:54:07 +0000