- 16 Apr, 2013 1 commit
-
-
Ingo Molnar authored
Merge branch 'uprobes/core' of git://git.kernel.org/pub/scm/linux/kernel/git/oleg/misc into perf/core Pull uprobes updates from Oleg Nesterov: - "uretprobes" - an optimization to uprobes, like kretprobes are an optimization to kprobes. "perf probe -x file sym%return" now works like kretprobes. - PowerPC fixes plus a couple of cleanups/optimizations in uprobes and trace_uprobes. Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
- 15 Apr, 2013 1 commit
-
-
Oleg Nesterov authored
perf_trace_buf_prepare() + perf_trace_buf_submit() make no sense if this task/CPU has no active counters. Change uprobe_perf_print() to return if hlist_empty(call->perf_events). Note: this is not uprobe-specific, we can change other users too. Signed-off-by:Oleg Nesterov <oleg@redhat.com>
-
- 13 Apr, 2013 20 commits
-
-
Oleg Nesterov authored
uprobe_perf_print() passes addr=ip to perf_trace_buf_submit() for no reason. This sets perf_sample_data->addr for PERF_SAMPLE_ADDR, we already have perf_sample_data->ip initialized if PERF_SAMPLE_IP. Signed-off-by:
Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Acked-by:
Srikar Dronamraju <srikar@linux.vnet.ibm.com>
-
Oleg Nesterov authored
Finally change create_trace_uprobe() to check if argv[0][0] == 'r' and pass the correct "is_ret" to alloc_trace_uprobe(). Signed-off-by:
Anton Arapov <anton@redhat.com>
-
Oleg Nesterov authored
Change probes_seq_show() and print_uprobe_event() to check is_ret_probe() and print the correct data. Signed-off-by:
-
Oleg Nesterov authored
Change uprobe_event_define_fields(), and __set_print_fmt() to check is_ret_probe() and use the appropriate format/fields. Signed-off-by:
-
Oleg Nesterov authored
Change uprobe_trace_print() and uprobe_perf_print() to check is_ret_probe() and fill ring_buffer_event accordingly. Also change uprobe_trace_func() and uprobe_perf_func() to not _print() if is_ret_probe() is true. Note that we keep ->handler() nontrivial even for uretprobe, we need this for filtering and for other potential extensions. Signed-off-by:
-
Oleg Nesterov authored
Create the new functions we need to support uretprobes, and change alloc_trace_uprobe() to initialize consumer.ret_handler if the new "is_ret" argument is true. Curently this argument is always false, so the new code is never called and is_ret_probe(tu) is false too. Signed-off-by:
-
Oleg Nesterov authored
Extract the output code from uprobe_trace_func() and uprobe_perf_func() into the new helpers, they will be used by ->ret_handler() too. We also add the unused "unsigned long func" argument in advance, to simplify the next changes. Signed-off-by:
-
Oleg Nesterov authored
struct uprobe_trace_entry_head has a single member for reporting, "unsigned long ip". If we want to support uretprobes we need to create another struct which has "func" and "ret_ip" and duplicate a lot of functions, like trace_kprobe.c does. To avoid this copy-and-paste horror we turn ->ip into ->vaddr[] and add couple of trivial helpers to calculate sizeof/data. This uglifies the code a bit, but this allows us to avoid a lot more complications later, when we add the support for ret-probes. Signed-off-by:
-
Oleg Nesterov authored
uprobe_trace_func() is never called with irqs or preemption disabled, no need to ask preempt_count() or local_save_flags(). Signed-off-by:
-
Oleg Nesterov authored
seq_print_ip_sym(ip) in print_uprobe_event() is pointless, kallsyms_lookup(ip) can not resolve a user-space address. Signed-off-by:
-
Oleg Nesterov authored
uprobe_trace_func() and uprobe_perf_func() do not need task_pt_regs(), we already have "struct pt_regs *regs". Signed-off-by:
-
Anton Arapov authored
add the uretprobe syntax and update an example Signed-off-by:
-
Anton Arapov authored
Enclose return probes implementation. Signed-off-by:
-
Anton Arapov authored
Unlike the kretprobes we can't trust userspace, thus must have protection from user space attacks. User-space have "unlimited" stack, and this patch limits the return probes nestedness as a simple remedy for it. Note that this implementation leaks return_instance on siglongjmp until exit()/exec(). The intention is to have KISS and bare minimum solution for the initial implementation in order to not complicate the uretprobes code. In the future we may come up with more sophisticated solution that remove this depth limitation. It is not easy task and lays beyond this patchset. Signed-off-by:
-
Anton Arapov authored
Uretprobe handlers are invoked when the trampoline is hit, on completion the trampoline is replaced with the saved return address and the uretprobe instance deleted. TODO: handle_trampoline() assumes that ->return_instances is always valid. We should teach it to handle longjmp() which can invalidate the pending return_instance's. This is nontrivial, we will try to do this in a separate series. Signed-off-by:
-
Anton Arapov authored
When a uprobe with return probe consumer is hit, prepare_uretprobe() function is invoked. It creates return_instance, hijacks return address and replaces it with the trampoline. * Return instances are kept as stack per uprobed task. * Return instance is chained, when the original return address is trampoline's page vaddr (e.g. recursive call of the probed function). Signed-off-by:
-
Anton Arapov authored
Hijack the return address and replace it with a trampoline address. PowerPC implementation. Signed-off-by:
Ananth N Mavinakayanahalli <ananth@in.ibm.com> Acked-by:
-
Anton Arapov authored
Hijack the return address and replace it with a trampoline address. Signed-off-by:
-
Anton Arapov authored
Allocate trampoline page, as the very first one in uprobed task xol area, and fill it with breakpoint opcode. Also introduce get_trampoline_vaddr() helper, to wrap the trampoline address extraction from area->vaddr. That removes confusion and eases the debug experience in case ->vaddr notion will be changed. Signed-off-by:
-
Anton Arapov authored
Enclose return probes implementation, introduce ->ret_handler() and update existing code to rely on ->handler() *and* ->ret_handler() for uprobe and uretprobe respectively. Signed-off-by:
-
- 10 Apr, 2013 1 commit
-
-
Andi Kleen authored
Add CYCLE_ACTIVITY.CYCLES_NO_DISPATCH/CYCLES_L1D_PENDING constraints. These recently documented events have restrictions to counter 0-3 and counter 2 respectively. The perf scheduler needs to know that to schedule them correctly. IvyBridge already has the necessary constraints. Signed-off-by:
Andi Kleen <ak@linux.intel.com> Cc: a.p.zijlstra@chello.nl Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/r/1362784968-12542-1-git-send-email-andi@firstfloor.orgSigned-off-by:
-
- 08 Apr, 2013 4 commits
-
-
Masami Hiramatsu authored
Return an error from __copy_instruction() and use printk() to give us a more productive message, since this is just an error case which we can handle and also the BUG_ON() never tells us why and what happened. This is related to the following bug-report: https://bugzilla.redhat.com/show_bug.cgi?id=910649Signed-off-by:
-
Masami Hiramatsu authored
Currently, __kprobes is defined in linux/kprobes.h which is too big to be included in small or basic headers that want to make use of this simple attribute. So move __kprobes definition into linux/compiler.h in which other compiler attributes are defined. Signed-off-by:
-
Stephane Eranian authored
This small patch fixes a mistake in the comments for the PERF_MEM_LVL_* events. The L2, L3 bits simply represent cache levels, not hits or misses. That is encoded in PERF_MEM_LVL_MISS/PERF_MEM_LVL_HIT. Signed-off-by:
Stephane Eranian <eranian@google.com> Cc: peterz@infradead.org Cc: jolsa@redhat.com Link: http://lkml.kernel.org/r/20130405144941.GA30503@quadSigned-off-by:
-
Ingo Molnar authored
Merge branch 'for-tip' of git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile into perf/core Pull IBM zEnterprise EC12 support patchlet from Robert Richter. Signed-off-by:
-
- 06 Apr, 2013 1 commit
-
-
Andreas Krebbel authored
This patch adds support for the latest release of the IBM mainframe series - the IBM zEnterprise EC12 (zEC12). The CPU measurement facility didn't change. So only the new CPU type has to be tolerated. Signed-off-by:
Andreas Krebbel <krebbel@linux.vnet.ibm.com> Signed-off-by:
Robert Richter <rric@kernel.org>
-
- 04 Apr, 2013 9 commits
-
-
Oleg Nesterov authored
Change write_opcode() to use copy_highpage() + copy_to_page() and simplify the code. Signed-off-by:
-
Oleg Nesterov authored
Extract the kmap_atomic/memcpy/kunmap_atomic code from xol_get_insn_slot() into the new simple helper, copy_to_page(). It will have more users soon. Signed-off-by:
-
Oleg Nesterov authored
__copy_insn(filp) can only be called after valid_vma() returns T, vma->vm_file passed as "filp" can not be NULL. Signed-off-by:
-
Oleg Nesterov authored
Change __copy_insn() to use copy_from_page() and simplify the code. Signed-off-by:
-
Oleg Nesterov authored
No functional changes. Rename copy_opcode() into copy_from_page() and add the new "int len" argument to make it more more generic for the new users. Signed-off-by:
-
Ananth N Mavinakayanahalli authored
prepare_uprobe() already checks if the underlying unstruction (on file) is a trap variant. We don't need to check this again. Signed-off-by:
-
Ananth N Mavinakayanahalli authored
Powerpc has many trap variants that could be used by entities like gdb. Currently, running gdb on a program being traced by uprobes causes an endless loop since uprobes doesn't understand that the trap was inserted by some other entity and a SIGTRAP needs to be delivered. Teach uprobes to ignore breakpoints that do not belong to it. Signed-off-by:
-
Ananth N Mavinakayanahalli authored
Some architectures like powerpc have multiple variants of the trap instruction. Introduce an additional helper is_trap_insn() for run-time handling of non-uprobe traps on such architectures. While there, change is_swbp_at_addr() to is_trap_at_addr() for reading clarity. With this change, the uprobe registration path will supercede any trap instruction inserted at the requested location, while taking care of delivering the SIGTRAP for cases where the trap notification came in for an address without a uprobe. See [1] for a more detailed explanation. [1] https://lists.ozlabs.org/pipermail/linuxppc-dev/2013-March/104771.html This change was suggested by Oleg Nesterov. Signed-off-by:
-
Oleg Nesterov authored
Cleanup. Now that we have f_inode/file_inode() we can use it instead of vm_file->f_mapping->host. This should not make any difference for uprobes, but in theory this change is more correct. We use this inode as a key, to compare it with uprobe->inode set by uprobe_register(inode), and the caller uses d_inode. Signed-off-by:
-
- 01 Apr, 2013 3 commits
-
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo: * Revert "perf sched: Handle PERF_RECORD_EXIT events" to get 'perf sched lat' back working. * We don't use Newt anymore, just plain libslang. * Kill a bunch of die() calls, from Namhyung Kim. * Add --no-demangle to report/top, from Namhyung Kim. * Fix dependency of the python binding wrt libtraceevent, from Naohiro Aota. * Introduce per core aggregation in 'perf stat', from Stephane Eranian. * Add memory profiling via PEBS, from Stephane Eranian. Signed-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by:
-
Arnaldo Carvalho de Melo authored
Initial motivation was to avoid the confusing exit when when '/' is pressed in non verbose mode, as specified in the help line searches are only available in verbose mode. Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-20xezxim2y4agmkx7f3sucll@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Now that the map browser shares the input routine with the hists browser, there is no need for using any libnewt routine, so remove all traces except for honouring NO_NEWT=1 on the makefile command line as an indication that TUI support is not needed, in fact it just sets NO_SLANG=1. Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-wae5o7xca9m52bj1re28jc5j@git.kernel.orgSigned-off-by:
-
