From a5f7a5bb3b7f28ba7e4fa246775b29a0e5537255 Mon Sep 17 00:00:00 2001 From: Jinke Han Date: Tue, 8 Sep 2026 15:37:42 +0800 Subject: x86/kprobes: Fix crash when probing CS CALL instructions When using eBPF to probe CS CALL instructions within a function, a crash can be triggered. The eBPF tool probes offset 257 of the __hrtimer_run_queues() function: <__hrtimer_run_queues+249>: nopl 0x0(%rax,%rax,1) <__hrtimer_run_queues+254>: mov %r14,%rdi <__hrtimer_run_queues+257>: cs call <__x86_indirect_thunk_r12> <__hrtimer_run_queues+263>: mov %eax,%r12d <__hrtimer_run_queues+266>: xchg %ax,%ax <__hrtimer_run_queues+268>: mov %r13,%rdi Which triggers this crash: BUG: unable to handle page fault for address: 00000000000f41c9 #PF: supervisor write access in kernel mode #PF: error_code(0x0002) - not-present page PGD 0 P4D 0 Oops: 0002 [#1] SMP NOPTI CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Tainted: P RIP: 0010:__hrtimer_run_queues+0x106/0x230 Note that __hrtimer_run_queues+0x106 is __hrtimer_run_queues+262, which is at the 6th byte of the above CS CALL instruction. Since the CS CALL instruction occupies 6 bytes, the exception occurred in the middle of that call instruction. The root cause is that when using eBPF tools to probe in the middle of a function, a kprobe with INT3 is used as the underlying implementation. During single-step emulation of the original CALL instruction, int3_emulate_call() assumes that the probed CALL instruction is 5 bytes long. However, the actual CS-prefixed CALL instruction occupies 6 bytes, so it constructs an incorrect exception return address. When the CPU returns from the kprobe handler, the next instruction to be executed is at the address of the last byte of that CS CALL instruction. Coincidentally, starting from that address, the CPU fetches and decodes a completely different instruction, which ultimately triggers a kernel crash. Fix the issue by using the actual instruction length obtained from the instruction decoder when constructing the exception return address, rather than relying on the hardcoded CALL_INSN_SIZE macro. [ mingo: Refined the changelog ] Fixes: 6256e668b7af ("x86/kprobes: Use int3 instead of debug trap for single-step") Suggested-by: Masami Hiramatsu (Google) Signed-off-by: Jinke Han Signed-off-by: Ingo Molnar Reviewed-by: Masami Hiramatsu (Google) Acked-by: Yafang Shao Acked-by: Borislav Petkov Cc: Peter Zijlstra Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000 --- Documentation/admin-guide/dynamic-debug-howto.rst | 392 ++++++++++++++++++++++ 1 file changed, 392 insertions(+) create mode 100644 Documentation/admin-guide/dynamic-debug-howto.rst (limited to 'Documentation/admin-guide/dynamic-debug-howto.rst') diff --git a/Documentation/admin-guide/dynamic-debug-howto.rst b/Documentation/admin-guide/dynamic-debug-howto.rst new file mode 100644 index 000000000..9c2f096ed --- /dev/null +++ b/Documentation/admin-guide/dynamic-debug-howto.rst @@ -0,0 +1,392 @@ +Dynamic debug ++++++++++++++ + + +Introduction +============ + +Dynamic debug allows you to dynamically enable/disable kernel +debug-print code to obtain additional kernel information. + +If ``/proc/dynamic_debug/control`` exists, your kernel has dynamic +debug. You'll need root access (sudo su) to use this. + +Dynamic debug provides: + + * a Catalog of all *prdbgs* in your kernel. + ``cat /proc/dynamic_debug/control`` to see them. + + * a Simple query/command language to alter *prdbgs* by selecting on + any combination of 0 or 1 of: + + - source filename + - function name + - line number (including ranges of line numbers) + - module name + - format string + - class name (as known/declared by each module) + +NOTE: To actually get the debug-print output on the console, you may +need to adjust the kernel ``loglevel=``, or use ``ignore_loglevel``. +Read about these kernel parameters in +Documentation/admin-guide/kernel-parameters.rst. + +Viewing Dynamic Debug Behaviour +=============================== + +You can view the currently configured behaviour in the *prdbg* catalog:: + + :#> head -n7 /proc/dynamic_debug/control + # filename:lineno [module]function flags format + init/main.c:1179 [main]initcall_blacklist =_ "blacklisting initcall %s\n" + init/main.c:1218 [main]initcall_blacklisted =_ "initcall %s blacklisted\n" + init/main.c:1424 [main]run_init_process =_ " with arguments:\n" + init/main.c:1426 [main]run_init_process =_ " %s\n" + init/main.c:1427 [main]run_init_process =_ " with environment:\n" + init/main.c:1429 [main]run_init_process =_ " %s\n" + +The 3rd space-delimited column shows the current flags, preceded by +a ``=`` for easy use with grep/cut. ``=p`` shows enabled callsites. + +Controlling dynamic debug Behaviour +=================================== + +The behaviour of *prdbg* sites are controlled by writing +query/commands to the control file. Example:: + + # grease the interface + :#> alias ddcmd='echo $* > /proc/dynamic_debug/control' + + :#> ddcmd '-p; module main func run* +p' + :#> grep =p /proc/dynamic_debug/control + init/main.c:1424 [main]run_init_process =p " with arguments:\n" + init/main.c:1426 [main]run_init_process =p " %s\n" + init/main.c:1427 [main]run_init_process =p " with environment:\n" + init/main.c:1429 [main]run_init_process =p " %s\n" + +Error messages go to console/syslog:: + + :#> ddcmd mode foo +p + dyndbg: unknown keyword "mode" + dyndbg: query parse failed + bash: echo: write error: Invalid argument + +If debugfs is also enabled and mounted, ``dynamic_debug/control`` is +also under the mount-dir, typically ``/sys/kernel/debug/``. + +Command Language Reference +========================== + +At the basic lexical level, a command is a sequence of words separated +by spaces or tabs. So these are all equivalent:: + + :#> ddcmd file svcsock.c line 1603 +p + :#> ddcmd "file svcsock.c line 1603 +p" + :#> ddcmd ' file svcsock.c line 1603 +p ' + +Command submissions are bounded by a write() system call. +Multiple commands can be written together, separated by ``;`` or ``\n``:: + + :#> ddcmd "func pnpacpi_get_resources +p; func pnp_assign_mem +p" + :#> ddcmd <<"EOC" + func pnpacpi_get_resources +p + func pnp_assign_mem +p + EOC + :#> cat query-batch-file > /proc/dynamic_debug/control + +You can also use wildcards in each query term. The match rule supports +``*`` (matches zero or more characters) and ``?`` (matches exactly one +character). For example, you can match all usb drivers:: + + :#> ddcmd file "drivers/usb/*" +p # "" to suppress shell expansion + +Syntactically, a command is pairs of keyword values, followed by a +flags change or setting:: + + command ::= match-spec* flags-spec + +The match-spec's select *prdbgs* from the catalog, upon which to apply +the flags-spec, all constraints are ANDed together. An absent keyword +is the same as keyword "*". + +Note that since the match-spec can be empty, the flags are checked 1st, +then the pairs of keyword and value. Flag errs will hide keyword errs:: + + bash-5.2# ddcmd mod bar +foo + dyndbg: read 13 bytes from userspace + dyndbg: query 0: "mod bar +foo" mod:* + dyndbg: unknown flag 'o' + dyndbg: flags parse failed + dyndbg: processed 1 queries, with 0 matches, 1 errs + +So a match-spec is a keyword, which selects the attribute of the +callsite to be compared, and a value to compare against. Possible +keywords are:: + + match-spec ::= 'func' string | + 'file' string | + 'module' string | + 'format' string | + 'class' string | + 'line' line-range + + line-range ::= lineno | + '-'lineno | + lineno'-' | + lineno'-'lineno + + lineno ::= unsigned-int + +.. note:: + + ``line-range`` cannot contain space, e.g. + "1-30" is valid range but "1 - 30" is not. + + +The meanings of each keyword are: + +func + The given string is compared against the function name + of each callsite. Example:: + + func svc_tcp_accept + func *recv* # in rfcomm, bluetooth, ping, tcp + +file + The given string is compared against either the src-root relative + pathname, or the basename of the source file of each callsite. + Examples:: + + file svcsock.c + file kernel/freezer.c # ie column 1 of control file + file drivers/usb/* # all callsites under it + file inode.c:start_* # parse :tail as a func (above) + file inode.c:1-100 # parse :tail as a line-range (above) + +module + The given string is compared against the module name + of each callsite. The module name is the string as + seen in ``lsmod``, i.e. without the directory or the ``.ko`` + suffix and with ``-`` changed to ``_``. Examples:: + + module sunrpc + module nfsd + module drm* # both drm, drm_kms_helper + +format + The given string is searched for in the dynamic debug format + string. Note that the string does not need to match the + entire format, only some part. Whitespace and other + special characters can be escaped using C octal character + escape ``\ooo`` notation, e.g. the space character is ``\040``. + Alternatively, the string can be enclosed in double quote + characters (``"``) or single quote characters (``'``). + Examples:: + + format svcrdma: // many of the NFS/RDMA server pr_debugs + format readahead // some pr_debugs in the readahead cache + format nfsd:\040SETATTR // one way to match a format with whitespace + format "nfsd: SETATTR" // a neater way to match a format with whitespace + format 'nfsd: SETATTR' // yet another way to match a format with whitespace + +class + The given class_name is validated against each module, which may + have declared a list of known class_names. If the class_name is + found for a module, callsite & class matching and adjustment + proceeds. Examples:: + + class DRM_UT_KMS # a DRM.debug category + class JUNK # silent non-match + // class TLD_* # NOTICE: no wildcard in class names + +line + The given line number or range of line numbers is compared + against the line number of each ``pr_debug()`` callsite. A single + line number matches the callsite line number exactly. A + range of line numbers matches any callsite between the first + and last line number inclusive. An empty first number means + the first line in the file, an empty last line number means the + last line number in the file. Examples:: + + line 1603 // exactly line 1603 + line 1600-1605 // the six lines from line 1600 to line 1605 + line -1605 // the 1605 lines from line 1 to line 1605 + line 1600- // all lines from line 1600 to the end of the file + +The flags specification comprises a change operation followed +by one or more flag characters. The change operation is one +of the characters:: + + - remove the given flags + + add the given flags + = set the flags to the given flags + +The flags are:: + + p enables the pr_debug() callsite. + _ enables no flags. + + Decorator flags add to the message-prefix, in order: + t Include thread ID, or + m Include module name + f Include the function name + s Include the source file name + l Include line number + d Include call trace + +For ``print_hex_dump_debug()`` and ``print_hex_dump_bytes()``, only +the ``p`` flag has meaning, other flags are ignored. + +Note the regexp ``^[-+=][fslmptd_]+$`` matches a flags specification. +To clear all flags at once, use ``=_`` or ``-fslmptd``. + + +Debug messages during Boot Process +================================== + +To activate debug messages for core code and built-in modules during +the boot process, even before userspace and debugfs exists, use +``dyndbg="QUERY"`` or ``module.dyndbg="QUERY"``. QUERY follows +the syntax described above, but must not exceed 1023 characters. Your +bootloader may impose lower limits. + +These ``dyndbg`` params are processed just after the ddebug tables are +processed, as part of the early_initcall. Thus you can enable debug +messages in all code run after this early_initcall via this boot +parameter. + +On an x86 system for example ACPI enablement is a subsys_initcall and:: + + dyndbg="file ec.c +p" + +will show early Embedded Controller transactions during ACPI setup if +your machine (typically a laptop) has an Embedded Controller. +PCI (or other devices) initialization also is a hot candidate for using +this boot parameter for debugging purposes. + +If ``foo`` module is not built-in, ``foo.dyndbg`` will still be processed at +boot time, without effect, but will be reprocessed when module is +loaded later. Bare ``dyndbg=`` is only processed at boot. + + +Debug Messages at Module Initialization Time +============================================ + +When ``modprobe foo`` is called, modprobe scans ``/proc/cmdline`` for +``foo.params``, strips ``foo.``, and passes them to the kernel along with +params given in modprobe args or ``/etc/modprobe.d/*.conf`` files, +in the following order: + +1. parameters given via ``/etc/modprobe.d/*.conf``:: + + options foo dyndbg=+pt + options foo dyndbg # defaults to +p + +2. ``foo.dyndbg`` as given in boot args, ``foo.`` is stripped and passed:: + + foo.dyndbg=" func bar +p; func buz +mp" + +3. args to modprobe:: + + modprobe foo dyndbg==pmf # override previous settings + +These ``dyndbg`` queries are applied in order, with last having final say. +This allows boot args to override or modify those from ``/etc/modprobe.d`` +(sensible, since 1 is system wide, 2 is kernel or boot specific), and +modprobe args to override both. + +In the ``foo.dyndbg="QUERY"`` form, the query must exclude ``module foo``. +``foo`` is extracted from the param-name, and applied to each query in +``QUERY``, and only 1 match-spec of each type is allowed. + +The ``dyndbg`` option is a "fake" module parameter, which means: + +- modules do not need to define it explicitly +- every module gets it tacitly, whether they use pr_debug or not +- it doesn't appear in ``/sys/module/$module/parameters/`` + To see it, grep the control file, or inspect ``/proc/cmdline.`` + +For ``CONFIG_DYNAMIC_DEBUG`` kernels, any settings given at boot-time (or +enabled by ``-DDEBUG`` flag during compilation) can be disabled later via +the debugfs interface if the debug messages are no longer needed:: + + echo "module module_name -p" > /proc/dynamic_debug/control + +Examples +======== + +:: + + // enable the message at line 1603 of file svcsock.c + :#> ddcmd 'file svcsock.c line 1603 +p' + + // enable all the messages in file svcsock.c + :#> ddcmd 'file svcsock.c +p' + + // enable all the messages in the NFS server module + :#> ddcmd 'module nfsd +p' + + // enable all 12 messages in the function svc_process() + :#> ddcmd 'func svc_process +p' + + // disable all 12 messages in the function svc_process() + :#> ddcmd 'func svc_process -p' + + // enable messages for NFS calls READ, READLINK, READDIR and READDIR+. + :#> ddcmd 'format "nfsd: READ" +p' + + // enable messages in files of which the paths include string "usb" + :#> ddcmd 'file *usb* +p' + + // enable all messages + :#> ddcmd '+p' + + // add module, function to all enabled messages + :#> ddcmd '+mf' + + // boot-args example, with newlines and comments for readability + Kernel command line: ... + // see what's going on in dyndbg=value processing + dynamic_debug.verbose=3 + // enable pr_debugs in the btrfs module (can be builtin or loadable) + btrfs.dyndbg="+p" + // enable pr_debugs in all files under init/ + // and the function parse_one, #cmt is stripped + dyndbg="file init/* +p #cmt ; func parse_one +p" + // enable pr_debugs in 2 functions in a module loaded later + pc87360.dyndbg="func pc87360_init_device +p; func pc87360_find +p" + +Kernel Configuration +==================== + +Dynamic Debug is enabled via kernel config items:: + + CONFIG_DYNAMIC_DEBUG=y # build catalog, enables CORE + CONFIG_DYNAMIC_DEBUG_CORE=y # enable mechanics only, skip catalog + +If you do not want to enable dynamic debug globally (i.e. in some embedded +system), you may set ``CONFIG_DYNAMIC_DEBUG_CORE`` as basic support of dynamic +debug and add ``ccflags := -DDYNAMIC_DEBUG_MODULE`` into the Makefile of any +modules which you'd like to dynamically debug later. + + +Kernel *prdbg* API +================== + +The following functions are cataloged and controllable when dynamic +debug is enabled:: + + pr_debug() + dev_dbg() + print_hex_dump_debug() + print_hex_dump_bytes() + +Otherwise, they are off by default; ``ccflags += -DDEBUG`` or +``#define DEBUG`` in a source file will enable them appropriately. + +If ``CONFIG_DYNAMIC_DEBUG`` is not set, ``print_hex_dump_debug()`` is +just a shortcut for ``print_hex_dump(KERN_DEBUG)``. + +For ``print_hex_dump_debug()``/``print_hex_dump_bytes()``, format string is +its ``prefix_str`` argument, if it is constant string; or ``hexdump`` +in case ``prefix_str`` is built dynamically. -- cgit v1.3.1