Michael Jeanson [Mon, 26 Oct 2020 18:11:17 +0000 (14:11 -0400)]
fix: KVM: x86: Add intr/vectoring info and error code to kvm_exit tracepoint (v5.10)
See upstream commit :
commit
235ba74f008d2e0936b29f77f68d4e2f73ffd24a
Author: Sean Christopherson <sean.j.christopherson@intel.com>
Date: Wed Sep 23 13:13:46 2020 -0700
KVM: x86: Add intr/vectoring info and error code to kvm_exit tracepoint
Extend the kvm_exit tracepoint to align it with kvm_nested_vmexit in
terms of what information is captured. On SVM, add interrupt info and
error code, while on VMX it add IDT vectoring and error code. This
sets the stage for macrofying the kvm_exit tracepoint definition so that
it can be reused for kvm_nested_vmexit without loss of information.
Opportunistically stuff a zero for VM_EXIT_INTR_INFO if the VM-Enter
failed, as the field is guaranteed to be invalid. Note, it'd be
possible to further filter the interrupt/exception fields based on the
VM-Exit reason, but the helper is intended only for tracepoints, i.e.
an extra VMREAD or two is a non-issue, the failed VM-Enter case is just
low hanging fruit.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I638fa29ef7d8bb432de42a33f9ae4db43259b915
Michael Jeanson [Mon, 26 Oct 2020 21:03:23 +0000 (17:03 -0400)]
fix: ext4: fast commit recovery path (v5.10)
See upstream commit :
commit
8016e29f4362e285f0f7e38fadc61a5b7bdfdfa2
Author: Harshad Shirwadkar <harshadshirwadkar@gmail.com>
Date: Thu Oct 15 13:37:59 2020 -0700
ext4: fast commit recovery path
This patch adds fast commit recovery path support for Ext4 file
system. We add several helper functions that are similar in spirit to
e2fsprogs journal recovery path handlers. Example of such functions
include - a simple block allocator, idempotent block bitmap update
function etc. Using these routines and the fast commit log in the fast
commit area, the recovery path (ext4_fc_replay()) performs fast commit
log recovery.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ia65cf44e108f2df0b458f0d335f33a8f18f50baa
Michael Jeanson [Tue, 27 Oct 2020 16:10:05 +0000 (12:10 -0400)]
fix: btrfs: make ordered extent tracepoint take btrfs_inode (v5.10)
See upstream commit :
commit
acbf1dd0fcbd10c67826a19958f55a053b32f532
Author: Nikolay Borisov <nborisov@suse.com>
Date: Mon Aug 31 14:42:40 2020 +0300
btrfs: make ordered extent tracepoint take btrfs_inode
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I096d0801ffe0ad826cfe414cdd1c0857cbd2b624
Michael Jeanson [Tue, 27 Oct 2020 15:42:23 +0000 (11:42 -0400)]
fix: btrfs: tracepoints: output proper root owner for trace_find_free_extent() (v5.10)
See upstream commit :
commit
437490fed3b0c9ae21af8f70e0f338d34560842b
Author: Qu Wenruo <wqu@suse.com>
Date: Tue Jul 28 09:42:49 2020 +0800
btrfs: tracepoints: output proper root owner for trace_find_free_extent()
The current trace event always output result like this:
find_free_extent: root=2(EXTENT_TREE) len=16384 empty_size=0 flags=4(METADATA)
find_free_extent: root=2(EXTENT_TREE) len=16384 empty_size=0 flags=4(METADATA)
find_free_extent: root=2(EXTENT_TREE) len=8192 empty_size=0 flags=1(DATA)
find_free_extent: root=2(EXTENT_TREE) len=8192 empty_size=0 flags=1(DATA)
find_free_extent: root=2(EXTENT_TREE) len=4096 empty_size=0 flags=1(DATA)
find_free_extent: root=2(EXTENT_TREE) len=4096 empty_size=0 flags=1(DATA)
T's saying we're allocating data extent for EXTENT tree, which is not
even possible.
It's because we always use EXTENT tree as the owner for
trace_find_free_extent() without using the @root from
btrfs_reserve_extent().
This patch will change the parameter to use proper @root for
trace_find_free_extent():
Now it looks much better:
find_free_extent: root=5(FS_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)
find_free_extent: root=5(FS_TREE) len=8192 empty_size=0 flags=1(DATA)
find_free_extent: root=5(FS_TREE) len=16384 empty_size=0 flags=1(DATA)
find_free_extent: root=5(FS_TREE) len=4096 empty_size=0 flags=1(DATA)
find_free_extent: root=5(FS_TREE) len=8192 empty_size=0 flags=1(DATA)
find_free_extent: root=5(FS_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)
find_free_extent: root=7(CSUM_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)
find_free_extent: root=2(EXTENT_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)
find_free_extent: root=1(ROOT_TREE) len=16384 empty_size=0 flags=36(METADATA|DUP)
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1d674064d29b31417e2acffdeb735f5052a87032
Michael Jeanson [Mon, 26 Oct 2020 17:41:02 +0000 (13:41 -0400)]
fix: objtool: Rename frame.h -> objtool.h (v5.10)
See upstream commit :
commit
00089c048eb4a8250325efb32a2724fd0da68cce
Author: Julien Thierry <jthierry@redhat.com>
Date: Fri Sep 4 16:30:25 2020 +0100
objtool: Rename frame.h -> objtool.h
Header frame.h is getting more code annotations to help objtool analyze
object files.
Rename the file to objtool.h.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ic2283161bebcbf1e33b72805eb4d2628f4ae3e89
Mathieu Desnoyers [Thu, 19 Nov 2020 16:03:17 +0000 (11:03 -0500)]
Fix: ressource leak in id tracker
Memory leak found by Coverity:
CID
1412251 (#2 of 2): Resource leak (RESOURCE_LEAK)
21. leaked_storage: Variable head going out of scope leaks the storage it points to.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Mon, 5 Oct 2020 19:31:42 +0000 (15:31 -0400)]
fix: strncpy equals destination size warning
Some versions of GCC when called with -Wstringop-truncation will warn
when doing a copy of the same size as the destination buffer with
strncpy :
‘strncpy’ specified bound 256 equals destination size [-Werror=stringop-truncation]
Since we unconditionally write '\0' in the last byte, reduce the copy
size by one.
Change-Id: Idb907c9550817a06fc0dffc489740f63d440e7d4
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Mathieu Desnoyers [Mon, 5 Oct 2020 19:29:51 +0000 (15:29 -0400)]
Update Changelog for Version 2.12.3
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 5 Oct 2020 16:01:37 +0000 (12:01 -0400)]
Cleanup: lttng-syscalls: silence warning about uninitialized bitmap variable
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Fri, 2 Oct 2020 20:10:05 +0000 (16:10 -0400)]
fix: backport 'Add 'kernel_read' wrapper for kernels < v4.14'
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I3b558a6a4b850054d5786bdf99e0849091c83eae
Michael Jeanson [Fri, 2 Oct 2020 17:03:34 +0000 (13:03 -0400)]
Add 'kernel_read' wrapper for kernels < v4.14
See upstream commit:
commit
bdd1d2d3d251c65b74ac4493e08db18971c09240
Author: Christoph Hellwig <hch@lst.de>
Date: Fri Sep 1 17:39:13 2017 +0200
fs: fix kernel_read prototype
Use proper ssize_t and size_t types for the return value and count
argument, move the offset last and make it an in/out argument like
all other read/write helpers, and make the buf argument a void pointer
to get rid of lots of casts in the callers.
Change-Id: I825c3fcbcc17e9b46e2a661fadc66b52a94eb2da
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 2 Oct 2020 16:22:27 +0000 (12:22 -0400)]
Version 2.12.3
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Thu, 24 Sep 2020 19:38:35 +0000 (15:38 -0400)]
fix: Use 'kernel_read' to read from procfs
Use the 'kernel_read' helper to read files in procfs, it's present in
the kernel since the 2.6 series and does the right thing on kernels that
require the set_fs dance and newer one which don't.
Change-Id: I1a53fda379e0bb9acc79331626925bbdba63d727
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Fri, 25 Sep 2020 20:05:00 +0000 (16:05 -0400)]
fix: don't allow userspace copy to read kernel memory
This patch fixes a security issue which allows the root user to read
arbitrary kernel memory. Considering the security model used in LTTng
userspace tooling for kernel tracing, this bug also allows members of
the 'tracing' group to read arbitrary kernel memory.
Calls to __copy_from_user_inatomic() where wrongly enclosed in
set_fs(KERNEL_DS) defeating the access_ok() calls and allowing to read
from kernel memory if a kernel address is provided.
Remove all set_fs() calls around __copy_from_user_inatomic().
As a side effect this will allow us to support v5.10 which should remove
set_fs().
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I35e4562c835217352c012ed96a7b8f93e941381e
Michael Jeanson [Fri, 25 Sep 2020 15:23:58 +0000 (11:23 -0400)]
fix: Add a 1MB limit to lttng_strlen_user_inatomic
The previous implementation was unbounded which could result in long
loops with preemption turned off.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I85afcd879258735bb2e7502f6016fcb2d3974cf7
Michael Jeanson [Wed, 23 Sep 2020 18:42:18 +0000 (14:42 -0400)]
fix: Adjust ranges for Ubuntu 4.15.0-119 kernel
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ie32f70f810c8fc756fbd31ab129aeb35500790f7
Michael Jeanson [Wed, 16 Sep 2020 19:16:17 +0000 (15:16 -0400)]
fix: Adjust ranges for Ubuntu HWE 5.0 kernels
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I36f2c3485dcc6ccb74ea86a7ce66fcb1662d060b
Mathieu Desnoyers [Tue, 28 Jan 2020 21:02:44 +0000 (16:02 -0500)]
Fix: system call filter table
The system call filter table has effectively been unused for a long
time due to system call name prefix mismatch. This means the overhead of
selective system call tracing was larger than it should have been because
the event payload preparation would be done for all system calls as soon
as a single system call is traced.
However, fixing this underlying issue unearths several issues that crept
unnoticed when the "enabler" concept was introduced (after the original
implementation of the system call filter table).
Here is a list of the issues which are resolved here:
- Split lttng_syscalls_unregister into an unregister and destroy
function, thus awaiting for a grace period (and therefore quiescence
of the users) after unregistering the system call tracepoints before
freeing the system call filter data structures. This effectively fixes
a use-after-free.
- The state for enabling "all" system calls vs enabling specific system
calls (and sequences of enable-disable) was incorrect with respect to
the "enablers" semantic. This is solved by always tracking the
bitmap of enabled system calls, and keeping this bitmap even when
enabling all system calls. The sc_filter is now always allocated
before system call tracing is registered to tracepoints, which means
it does not need to be RCU dereferenced anymore.
Padding fields in the ABI are reserved to select whether to:
- Trace either native or compat system call (or both, which is the
behavior currently implemented),
- Trace either system call entry or exit (or both, which is the
behavior currently implemented),
- Select the system call to trace by name (behavior currently
implemented) or by system call number,
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Fri, 4 Sep 2020 15:52:51 +0000 (11:52 -0400)]
fix: version ranges for ext4_discard_preallocations and writeback_queue_io
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Id4fa53cb2e713cbda651e1a75deed91013115592
Michael Jeanson [Mon, 31 Aug 2020 18:16:01 +0000 (14:16 -0400)]
fix: writeback: Fix sync livelock due to b_dirty_time processing (v5.9)
See upstream commit:
commit
f9cae926f35e8230330f28c7b743ad088611a8de
Author: Jan Kara <jack@suse.cz>
Date: Fri May 29 16:08:58 2020 +0200
writeback: Fix sync livelock due to b_dirty_time processing
When we are processing writeback for sync(2), move_expired_inodes()
didn't set any inode expiry value (older_than_this). This can result in
writeback never completing if there's steady stream of inodes added to
b_dirty_time list as writeback rechecks dirty lists after each writeback
round whether there's more work to be done. Fix the problem by using
sync(2) start time is inode expiry value when processing b_dirty_time
list similarly as for ordinarily dirtied inodes. This requires some
refactoring of older_than_this handling which simplifies the code
noticeably as a bonus.
Change-Id: I8b894b13ccc14d9b8983ee4c2810a927c319560b
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Mon, 31 Aug 2020 15:41:38 +0000 (11:41 -0400)]
fix: writeback: Drop I_DIRTY_TIME_EXPIRE (v5.9)
See upstream commit:
commit
5fcd57505c002efc5823a7355e21f48dd02d5a51
Author: Jan Kara <jack@suse.cz>
Date: Fri May 29 16:24:43 2020 +0200
writeback: Drop I_DIRTY_TIME_EXPIRE
The only use of I_DIRTY_TIME_EXPIRE is to detect in
__writeback_single_inode() that inode got there because flush worker
decided it's time to writeback the dirty inode time stamps (either
because we are syncing or because of age). However we can detect this
directly in __writeback_single_inode() and there's no need for the
strange propagation with I_DIRTY_TIME_EXPIRE flag.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I92e37c2ff3ec36d431e8f9de5c8e37c5a2da55ea
Michael Jeanson [Tue, 25 Aug 2020 14:56:29 +0000 (10:56 -0400)]
fix: removal of [smp_]read_barrier_depends (v5.9)
See upstream commits:
commit
76ebbe78f7390aee075a7f3768af197ded1bdfbb
Author: Will Deacon <will@kernel.org>
Date: Tue Oct 24 11:22:47 2017 +0100
locking/barriers: Add implicit smp_read_barrier_depends() to READ_ONCE()
In preparation for the removal of lockless_dereference(), which is the
same as READ_ONCE() on all architectures other than Alpha, add an
implicit smp_read_barrier_depends() to READ_ONCE() so that it can be
used to head dependency chains on all architectures.
commit
76ebbe78f7390aee075a7f3768af197ded1bdfbb
Author: Will Deacon <will.deacon@arm.com>
Date: Tue Oct 24 11:22:47 2017 +0100
locking/barriers: Add implicit smp_read_barrier_depends() to READ_ONCE()
In preparation for the removal of lockless_dereference(), which is the
same as READ_ONCE() on all architectures other than Alpha, add an
implicit smp_read_barrier_depends() to READ_ONCE() so that it can be
used to head dependency chains on all architectures.
Change-Id: Ife8880bd9378dca2972da8838f40fc35ccdfaaac
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Mon, 24 Aug 2020 19:37:50 +0000 (15:37 -0400)]
fix: ext4: indicate via a block bitmap read is prefetched… (v5.9)
See upstream commit:
commit
ab74c7b23f3770935016e3eb3ecdf1e42b73efaa
Author: Theodore Ts'o <tytso@mit.edu>
Date: Wed Jul 15 11:48:55 2020 -0400
ext4: indicate via a block bitmap read is prefetched via a tracepoint
Modify the ext4_read_block_bitmap_load tracepoint so that it tells us
whether a block bitmap is being prefetched.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I0e5e2c5b8004223d0928235c092449ee16a940e1
Michael Jeanson [Mon, 24 Aug 2020 19:26:04 +0000 (15:26 -0400)]
fix: ext4: limit the length of per-inode prealloc list (v5.9)
See upstream commit:
commit
27bc446e2def38db3244a6eb4bb1d6312936610a
Author: brookxu <brookxu.cn@gmail.com>
Date: Mon Aug 17 15:36:15 2020 +0800
ext4: limit the length of per-inode prealloc list
In the scenario of writing sparse files, the per-inode prealloc list may
be very long, resulting in high overhead for ext4_mb_use_preallocated().
To circumvent this problem, we limit the maximum length of per-inode
prealloc list to 512 and allow users to modify it.
After patching, we observed that the sys ratio of cpu has dropped, and
the system throughput has increased significantly. We created a process
to write the sparse file, and the running time of the process on the
fixed kernel was significantly reduced, as follows:
Running time on unfixed kernel:
[root@TENCENT64 ~]# time taskset 0x01 ./sparse /data1/sparce.dat
real 0m2.051s
user 0m0.008s
sys 0m2.026s
Running time on fixed kernel:
[root@TENCENT64 ~]# time taskset 0x01 ./sparse /data1/sparce.dat
real 0m0.471s
user 0m0.004s
sys 0m0.395s
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I5169cb24853d4da32e2862a6626f1f058689b053
Michael Jeanson [Mon, 10 Aug 2020 15:36:03 +0000 (11:36 -0400)]
fix: KVM: x86/mmu: Make kvm_mmu_page definition and accessor internal-only (v5.9)
commit
985ab2780164698ec6e7d73fad523d50449261dd
Author: Sean Christopherson <sean.j.christopherson@intel.com>
Date: Mon Jun 22 13:20:32 2020 -0700
KVM: x86/mmu: Make kvm_mmu_page definition and accessor internal-only
Make 'struct kvm_mmu_page' MMU-only, nothing outside of the MMU should
be poking into the gory details of shadow pages.
Change-Id: Ia5c1b9c49c2b00dad1d5b17c50c3dc730dafda20
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Mon, 10 Aug 2020 15:22:05 +0000 (11:22 -0400)]
fix: Move mmutrace.h into the mmu/ sub-directory (v5.9)
commit
33e3042dac6bcc33b80835f7d7b502b1d74c457c
Author: Sean Christopherson <sean.j.christopherson@intel.com>
Date: Mon Jun 22 13:20:29 2020 -0700
KVM: x86/mmu: Move mmu_audit.c and mmutrace.h into the mmu/ sub-directory
Move mmu_audit.c and mmutrace.h under mmu/ where they belong.
Change-Id: I582525ccca34e1e3bd62870364108a7d3e9df2e4
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Beniamin Sandu [Thu, 13 Aug 2020 13:24:39 +0000 (16:24 +0300)]
Kconfig: fix dependency issue when building in-tree without CONFIG_FTRACE
When building in-tree, one could disable CONFIG_FTRACE from kernel
config which will leave CONFIG_TRACEPOINTS selected by LTTNG modules,
but generate a lot of linker errors like below because it leaves out
other stuff, e.g.:
trace.c:(.text+0xd86b): undefined reference to `trace_event_buffer_reserve'
ld: trace.c:(.text+0xd8de): undefined reference to `trace_event_buffer_commit'
ld: trace.c:(.text+0xd926): undefined reference to `event_triggers_call'
ld: trace.c:(.text+0xd942): undefined reference to `trace_event_ignore_this_pid'
ld: net/mac80211/trace.o: in function `trace_event_raw_event_drv_tdls_cancel_channel_switch':
It appears to be caused by the fact that TRACE_EVENT macros in the Linux
kernel depend on the Ftrace ring buffer as soon as CONFIG_TRACEPOINTS is
enabled.
Steps to reproduce:
- Get a clone of an upstream stable kernel and use scripts/built-in.sh on it
- Configure a standard x86-64 build, enable built-in LTTNG but disable
CONFIG_FTRACE from Kernel Hacking-->Tracers using menuconfig
- Build will fail at linking stage
Signed-off-by: Beniamin Sandu <beniaminsandu@gmail.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Tue, 4 Aug 2020 13:46:24 +0000 (09:46 -0400)]
Version 2.12.2
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 13 Jul 2020 18:59:33 +0000 (14:59 -0400)]
Fix: Lock metadata cache on session destroy
commit
92143b2c5656 ("Fix: metadata stream leak, missing list removal and locking")
missed taking a lock protecting the metadata stream list iteration on
session destroy. This opens a race window between iteration and item
removal/free which triggers kernel OOPS.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jul 2020 15:15:40 +0000 (11:15 -0400)]
Fix: metadata stream leak, missing list removal and locking
The metadata stream is part of a list of metadata streams in the
metadata cache. Its addition to the list should be protected by
the metadata cache lock. It needs to be paired with protection
of list iteration with the same lock.
Removal from the list is entirely missing, and should be added
to lttng_metadata_ring_buffer_release (with proper locking).
This missing list removal was probably not causing issues because the
metadata stream structure was leaked: a kfree() is missing from
lttng_metadata_ring_buffer_release as well.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 10 Jul 2020 14:51:26 +0000 (10:51 -0400)]
Fix: coherent state not changed atomically with metadata written
commit
122c63cb4310 ("Fix: Implement RING_BUFFER_GET_NEXT_SUBBUF_METADATA_CHECK")
introduces a new ioctl which returns a flag indicating whether the
metadata is in consistent state at the end of the sub-buffer.
That commit is meant to address metadata consistency issues observable
in live sessions.
However, the "consistent" state is false as soon as a producer is
active (between an outermost metadata_begin/end pair). Unfortunately,
if the last "RING_BUFFER_GET_NEXT_SUBBUF_METADATA_CHECK" operation is
done between the last metadata printf and "end" of the transaction, the
last consistency state will be false, and the consumer daemon will never
send metadata to the relay daemon. This in turn causes a live viewer to
wait for metadata endlessly.
This issue can be reproduced by running lttng-tools:
tests/regression/tools/live/test_kernel
as root in a loop.
We observe two things:
1) the poll operation blocks when there is no more metadata to send,
which means there is no mean to unblock when the consistency state
changes back to "true" without producing additional metadata,
2) Even if (1) was fixed, the expectation from an ABI perspective is
that the "coherent" state is only populated when
RING_BUFFER_GET_NEXT_SUBBUF_METADATA_CHECK succeeds. Therefore,
there is no way to let user-space know about conherency transition
unless additional metadata is generated.
Fixing this requires to hold the metadata cache lock across the entire
production of a coherent metadata transaction. This simpler scheme is
possible because the metadata is generated in a reallocated memory area
and not directly into a ring buffer anymore. This was not the case in
earlier lttng-modules versions, when the metadata was generated directly
into a ring buffer, which explains why this simpler scheme was not
implemented.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 7 Jul 2020 18:18:37 +0000 (14:18 -0400)]
fix: include module.h for EXPORT_SYMBOL_GPL
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ic337e1eb375791ace08560555dd02b37cbefcf25
Michael Jeanson [Tue, 7 Jul 2020 17:50:15 +0000 (13:50 -0400)]
fix: __lttng_vmalloc_node_range const caller introduced in v3.6
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ib13cf03b5ab11830a8732318a12713720cf1b3e3
Michael Jeanson [Tue, 7 Jul 2020 18:07:01 +0000 (14:07 -0400)]
fix: version range for overflow_callback
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1b8f1d59552a1723d3f4ed74780a2b57d13d0e52
Michael Jeanson [Tue, 7 Jul 2020 17:00:10 +0000 (13:00 -0400)]
fix: global_dirty_limit was introduced in v3.1
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Id97dbb2d0181a45c45cfed36c4be8753cabac283
Michael Jeanson [Tue, 7 Jul 2020 16:21:54 +0000 (12:21 -0400)]
fix: wrapper_uprobe_unregister is a void function
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ib4438da02aac3defd1245324d1b48f400f806d58
Michael Jeanson [Tue, 7 Jul 2020 15:58:03 +0000 (11:58 -0400)]
fix: prior to v4.0, __vmalloc_node_range had no vm_flags param
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ib476e32d109298d9ca3e6b6ab7ac8f63c50fb09f
Michael Jeanson [Tue, 7 Jul 2020 15:15:39 +0000 (11:15 -0400)]
fix: vmalloc on v5.8 without KALLSYMS
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ic945dad92e78a5bc2895a969a10c527e1349decf
Michael Jeanson [Thu, 14 May 2020 17:47:35 +0000 (13:47 -0400)]
Detect missing symbols used with kallsyms_lookup at compile time
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I19a9a31c386196899517899d861fe63611272139
Michael Jeanson [Thu, 2 Jul 2020 16:06:42 +0000 (12:06 -0400)]
Use exported symbol bdevname() instead of disk_name()
bdevname() is a simple wrapper over disk_name() but has the honor to be
exported. Using it removes the need for a kallsym wrapper.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ic2b2233c4db7826175c68edea69751ddcb17a5e6
Michael Jeanson [Fri, 3 Jul 2020 14:46:12 +0000 (10:46 -0400)]
Add git-review config
Add .gitreview for contributors wishing to use gerrit for patch
reviews.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I663e66a433ddb645f580c4b9f885db9c3a08e02f
Michael Jeanson [Thu, 2 Jul 2020 15:21:42 +0000 (11:21 -0400)]
fix: mm: remove vmalloc_sync_(un)mappings() (v5.8)
See upstream commit:
commit
73f693c3a705756032c2863bfb37570276902d7d
Author: Joerg Roedel <jroedel@suse.de>
Date: Mon Jun 1 21:52:36 2020 -0700
mm: remove vmalloc_sync_(un)mappings()
These functions are not needed anymore because the vmalloc and ioremap
mappings are now synchronized when they are created or torn down.
Remove all callers and function definitions.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ifdefa35b25b4906cde407360e608b77e47cc3808
Michael Jeanson [Mon, 15 Jun 2020 15:12:24 +0000 (11:12 -0400)]
fix: mm/writeback: discard NR_UNSTABLE_NFS, use NR_WRITEBACK (v5.8)
See upstream commit:
commit
8d92890bd6b8502d6aee4b37430ae6444ade7a8c
Author: NeilBrown <neilb@suse.de>
Date: Mon Jun 1 21:48:21 2020 -0700
mm/writeback: discard NR_UNSTABLE_NFS, use NR_WRITEBACK instead
After an NFS page has been written it is considered "unstable" until a
COMMIT request succeeds. If the COMMIT fails, the page will be
re-written.
These "unstable" pages are currently accounted as "reclaimable", either
in WB_RECLAIMABLE, or in NR_UNSTABLE_NFS which is included in a
'reclaimable' count. This might have made sense when sending the COMMIT
required a separate action by the VFS/MM (e.g. releasepage() used to
send a COMMIT). However now that all writes generated by ->writepages()
will automatically be followed by a COMMIT (since commit
919e3bd9a875
("NFS: Ensure we commit after writeback is complete")) it makes more
sense to treat them as writeback pages.
So this patch removes NR_UNSTABLE_NFS and accounts unstable pages in
NR_WRITEBACK and WB_WRITEBACK.
A particular effect of this change is that when
wb_check_background_flush() calls wb_over_bg_threshold(), the latter
will report 'true' a lot less often as the 'unstable' pages are no
longer considered 'dirty' (as there is nothing that writeback can do
about them anyway).
Currently wb_check_background_flush() will trigger writeback to NFS even
when there are relatively few dirty pages (if there are lots of unstable
pages), this can result in small writes going to the server (10s of
Kilobytes rather than a Megabyte) which hurts throughput. With this
patch, there are fewer writes which are each larger on average.
Where the NR_UNSTABLE_NFS count was included in statistics
virtual-files, the entry is retained, but the value is hard-coded as
zero. static trace points and warning printks which mentioned this
counter no longer report it.
Change-Id: I18080ca62bc6c1cd7d6da4cb27cc1521fbdca5e1
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Mon, 15 Jun 2020 15:06:13 +0000 (11:06 -0400)]
fix: block: remove the error argument to the block_bio_complete (v5.8)
See upstream commit:
commit
d24de76af836260a99ca2ba281a937bd5bc55591
Author: Christoph Hellwig <hch@lst.de>
Date: Wed Jun 3 07:14:43 2020 +0200
block: remove the error argument to the block_bio_complete tracepoint
The status can be trivially derived from the bio itself. That also avoid
callers like NVMe to incorrectly pass a blk_status_t instead of the errno,
and the overhead of translating the blk_status_t to the errno in the I/O
completion fast path when no tracing is enabled.
Fixes: 35fe0d12c8a3 ("nvme: trace bio completion")
Change-Id: I8d1463184d79bfab418a1755bfc6a0200170fff3
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Mon, 15 Jun 2020 14:51:41 +0000 (10:51 -0400)]
fix: pipe_buf_operations rework (v5.8)
See upstream commits:
commit
c928f642c29a5ffb02e16f2430b42b876dde69de
Author: Christoph Hellwig <hch@lst.de>
Date: Wed May 20 17:58:16 2020 +0200
fs: rename pipe_buf ->steal to ->try_steal
And replace the arcane return value convention with a simple bool
where true means success and false means failure.
[AV: braino fix folded in]
commit
b8d9e7f2411b0744df2ec33e80d7698180fef21a
Author: Christoph Hellwig <hch@lst.de>
Date: Wed May 20 17:58:15 2020 +0200
fs: make the pipe_buf_operations ->confirm operation optional
Just return 0 for success if it is not present.
commit
76887c256744740d6121af9bc4aa787712a1f694
Author: Christoph Hellwig <hch@lst.de>
Date: Wed May 20 17:58:14 2020 +0200
fs: make the pipe_buf_operations ->steal operation optional
Just return 1 for failure if it is not present.
Change-Id: Ic185632202470db1eb5b012e95e793ff2cb26be7
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 3 Jun 2020 02:53:27 +0000 (22:53 -0400)]
Update lttng-tracer.h version
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 3 Jun 2020 02:50:40 +0000 (22:50 -0400)]
Version 2.12.1
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Fri, 24 Apr 2020 19:49:42 +0000 (15:49 -0400)]
Fix: Implement RING_BUFFER_GET_NEXT_SUBBUF_METADATA_CHECK
Get next metadata subbuffer, returning a flag indicating whether the
metadata is guaranteed to be in a consistent state at the end of this
sub-buffer (can be parsed).
This can be used by the consumer to know whether the metadata can be
parsed at the end of this sub-buffer, which is useful to distinguish
between errors and incomplete metadata in live tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Fri, 15 May 2020 19:12:53 +0000 (15:12 -0400)]
fix: vmalloc_sync_mappings was backported to v5.5.12
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ie554d9c956afc2a8e114fe41e4b3c225d8af40a1
Stefan Bader [Mon, 18 May 2020 14:03:16 +0000 (16:03 +0200)]
Update: Additional kernel ranges for vmalloc_sync_mappings
Some Ubuntu kernels cannot be directly mapped to an upstream stable
version. Define distro specific ranges for those (4.15, 5.0, 5.3).
Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Ovidiu Panait [Thu, 14 May 2020 11:27:17 +0000 (14:27 +0300)]
Update: Use vmalloc_sync_mappings for stable kernels
Starting from v5.4.28/v5.2.37/v4.19.113/v4.14.175/v4.9.218/v4.4.218, stable
kernel branches backported v5.6 upstream commit [1], causing the following
warnings:
...
[ 483.242037] LTTng: vmalloc_sync_all symbol lookup failed.
[ 483.257056] Page fault handler and NMI tracing might trigger faults.
...
Extend check for vmalloc_sync_mappings for stable kernels as well.
[1] https://github.com/torvalds/linux/commit/
763802b53a427ed3cbd419dbba255c414fdd9e7c
[ Edit: minor coding style fix by Mathieu Desnoyers. ]
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Ovidiu Panait [Thu, 14 May 2020 10:05:24 +0000 (13:05 +0300)]
Fix: Use vmalloc_sync_mappings on kernel 5.6 as well
Upstream commit [1], that got rid of vmalloc_sync_all and introduced
vmalloc_sync_mappings, is a v5.6 commit:
$ git tag --contains
763802b53a427ed3cbd419dbba255c414fdd9e7c
v5.6
v5.6-rc7
v5.7-rc1
v5.7-rc2
v5.7-rc3
Extend the LINUX_VERSION_CODE check to v5.6 to fix the following warnings:
...
[ 483.242037] LTTng: vmalloc_sync_all symbol lookup failed.
[ 483.257056] Page fault handler and NMI tracing might trigger faults.
...
[1] https://github.com/torvalds/linux/commit/
763802b53a427ed3cbd419dbba255c414fdd9e7c
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Wed, 6 May 2020 15:11:29 +0000 (11:11 -0400)]
fix: add missing guid_t type to wrapper
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I0de39c24a7925b580fabbdaa12dbe05c43cfcd98
Michael Jeanson [Wed, 6 May 2020 15:03:32 +0000 (11:03 -0400)]
Fix: missing wrapper rename to wrapper_vmalloc_sync_mappings
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Idf7082a980c5a604bfef5c69906678b5083a9bbf
Mathieu Desnoyers [Tue, 5 May 2020 17:38:31 +0000 (13:38 -0400)]
Update for kernel 5.7: use vmalloc_sync_mappings on kernels >= 5.7
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 4 May 2020 19:00:53 +0000 (15:00 -0400)]
Unbreak LTTng for kernel 5.7
Linux commit
0bd476e6c67190b5eb7b6e105c8db8ff61103281 ("kallsyms:
unexport kallsyms_lookup_name() and kallsyms_on_each_symbol()") breaks
LTTng-modules by removing symbols used by the LTTng-modules out-of-tree
tracer.
I pointed this out when the change was originally considered before the
5.7 merge window. This generated some discussion but it did not lead to
any concrete proposal to fix the issue. [1]
The commit has been merged in the 5.7 merge window. At that point, as
maintainer of LTTng, I immediately raised a flag about this issue,
proposing an alternative approach to solve this: expose the few symbols
needed by LTTng to GPL modules. This was NACKed on the ground that the
Linux kernel cannot export GPL symbols when there are no in-tree
users. [2]
Steven Rostedt has shown interest in merging LTTng-modules upstream.
LTTng-modules being LGPL, this is very much doable. I have prepared a
tree of LTTng-modules "for upstreaming" and sent it to him privately so
he can review it. Even if in an ideal scenario LTTng-modules is merged
for the following merge window, it leaves LTTng-modules broken on the
5.7 kernel.
In order to ensure that the LTTng-modules kernel tracer continues working
for my end users on kernels 5.7 onwards, as a very last resort, this is
with great reluctance that I created this fix for LTTng modules. It
basically uses kprobes to lookup the kallsyms_lookup_name symbol, and
continues using kallsyms_lookup_name as before.
Link: https://lore.kernel.org/r/20200302192811.n6o5645rsib44vco@localhost
Link: https://lore.kernel.org/r/20200409193543.18115-1-mathieu.desnoyers@efficios.com
Link: https://lwn.net/Articles/817988/
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
CC: Thomas Gleixner <tglx@linutronix.de>
CC: Will Deacon <will@kernel.org>
CC: akpm@linux-foundation.org
CC: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
CC: Masami Hiramatsu <mhiramat@kernel.org>
CC: rostedt@goodmis.org
CC: Alexei Starovoitov <ast@kernel.org>
Mathieu Desnoyers [Mon, 4 May 2020 18:52:13 +0000 (14:52 -0400)]
Move lttng wrappers into own module
Currently, we only pull the wrapper symbols into a single sub-module,
either:
lttng-tracer.o:
- wrapper/random.o
- wrapper/trace-clock.o
- wrapper/page_alloc.o
or
lttng-statedump.o:
- wrapper/irqdesc.o
- wrapper/fdtable.o
Because lttng-tracer depends on lttng-statedump, we cannot just put all
wrappers into lttng-tracer.o, because it would create a circular
dependency. This will be an issue if we introduce common wrappers which
are used in both lttng-tracer.o and in lttng-statedump.o.
Introduce a new lttng-wrapper.o to contain all wrapper symbols for all
lttng modules.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 13 Apr 2020 16:16:43 +0000 (12:16 -0400)]
Introduce lttng_guid_gen wrapper for kernels >= 5.7.0
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 13 Apr 2020 15:44:23 +0000 (11:44 -0400)]
instrumentation: update x86 kvm instrumentation for kernel >= 5.7.0
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 13 Apr 2020 15:38:48 +0000 (11:38 -0400)]
instrumentation: update mm_vmscan for kernel >= 5.7.0
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 8 Apr 2020 13:31:05 +0000 (09:31 -0400)]
Version 2.12.0
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Thu, 2 Apr 2020 18:08:36 +0000 (14:08 -0400)]
fix: uaccess wrapper for CentOS >= 4.18.0-147
Fixes: #1253
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I2a79c1c0e897a6148e60e5599949cd2778d09d50
Michael Jeanson [Thu, 2 Apr 2020 18:08:09 +0000 (14:08 -0400)]
fix: ext4 instrumentation for CentOS >= 4.18.0-147
Fixes: #1253
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1fd54af16fbb02cd4b3ab7fc7d9232708088f1fd
Michael Jeanson [Thu, 2 Apr 2020 18:07:47 +0000 (14:07 -0400)]
fix: signal instrumentation for CentOS >= 4.18.0-147
Fixes: #1253
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I280013402df6f14222fbb912cdf64d80af3ab265
Michael Jeanson [Thu, 2 Apr 2020 18:07:21 +0000 (14:07 -0400)]
fix: kvm instrumentation for CentOS >= 4.18.0-147
Fixes: #1253
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ide20ebf51bec503866ffc96dda3e0b09ebeb14d6
Michael Jeanson [Thu, 2 Apr 2020 18:06:17 +0000 (14:06 -0400)]
fix: rcu instrumentation for CentOS >= 4.18.0-80
Fixes: #1253
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1735d2caa7215ce94272aaaa98cbbc8f3a10743d
Michael Jeanson [Mon, 30 Mar 2020 21:43:16 +0000 (17:43 -0400)]
Fix: update kvm instrumentation for Ubuntu 5.3.0-45
This commit introduced in 5.3.0-43 was dropped in 5.3.0-45 and reintroduced
in 5.3.0-46:
commit
795f8a34f279e17c279bba46da10f15c5dd00264
Author: Sean Christopherson <sean.j.christopherson@intel.com>
Date: Fri Dec 6 15:57:14 2019 -0800
KVM: x86: Use gpa_t for cr2/gpa to fix TDP support on 32-bit KVM
BugLink: https://bugs.launchpad.net/bugs/1867051
[ Upstream commit
736c291c9f36b07f8889c61764c28edce20e715d ]
Fun times!
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ia5f1a4ba355f592f09e964038b6334ddb3ad5153
Mathieu Desnoyers [Fri, 27 Mar 2020 14:29:08 +0000 (10:29 -0400)]
Version 2.12.0-rc3
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 24 Mar 2020 18:20:48 +0000 (14:20 -0400)]
Fix: update kvm instrumentation for Ubuntu 5.3.0-43
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1ce5f9ebba997fcc4cfbae6901eed479e2e1a79e
Michael Jeanson [Mon, 23 Mar 2020 18:48:24 +0000 (14:48 -0400)]
Fix: update kvm instrumentation for Ubuntu 4.15.0-92
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: Ib367b9a0ce3846f45313906e710a9a6d644e3955
Mathieu Desnoyers [Tue, 10 Mar 2020 19:00:29 +0000 (15:00 -0400)]
statedump: introduce file_table_address
Currently the LTTng-modules statedump simply iterates over all processes
in the system and assumes all threads share the same file descriptor
table, which is only true if threads were created with clone
CLONE_FILES.
Directly invoking clone without the CLONE_FILES creates threads which
belong to the same process, but have their own file descriptor table.
Therefore, model-wise, we cannot assume that all threads in a process
have the same fd table content.
Add a new "file_table_address" field to the lttng_statedump_process_state
event, which dumps the address of the thread's struct files_struct
pointer. This pointer is guaranteed to never be re-used while we hold
the RCU read-side lock (so for the entire iteration over
processes/threads).
For the lttng_statedump_file_descriptor event, remove the "pid" field
(which is semantically inaccurate) and add a "file_table_address" field,
which contains the struct files_struct address of the file table
containing the file descriptor.
An optimization is performed to eliminate most duplcated file table
content by skipping file table dump if the same file table address is
encountered consecutively while iterating over a process' threads.
This introduces a semantic change to the statedump fields, and will
therefore be introduced in lttng-modules 2.12 onwards, not backported as
a fix.
Fixes: #1245
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 2 Mar 2020 16:26:39 +0000 (11:26 -0500)]
Remove lttng-ftrace integration
The lttng-ftrace integration (LTTNG_KERNEL_FUNCTION instrumentation
type) was unused for a while now. The "function" probing is actually
done with kprobes and kretprobes (LTTNG_KERNEL_KPROBE and
LTTNG_KERNEL_KRETPROBE).
Remove it so a use of kallsyms_lookup_name() can be removed as well.
Note that in the future we could add back this support by using
register_ftrace_function() which is exported to kernel modules, but
considering that we have not been using this code for a while,
just remove the implementation for now.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 2 Mar 2020 16:03:19 +0000 (11:03 -0500)]
Remove dependency on kallsyms for splice_to_pipe (kernel 4.2+)
Upstream commit
2b514574f7e88 "net: af_unix: implement splice for stream
af_unix sockets" exported the "splice_to_pipe" symbol, so use it to
remove a dependency on kallsyms_lookup_name().
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 2 Mar 2020 15:52:54 +0000 (10:52 -0500)]
Remove dependency on kallsyms for irq_to_desc (kernel 3.4+)
Upstream commit
3911ff30f5d "genirq: export handle_edge_irq() and
irq_to_desc()" exported the irq_to_desc symbol, so use it to remove a
dependency on kallsyms_lookup_name().
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 2 Mar 2020 15:40:26 +0000 (10:40 -0500)]
Remove work-around for signed tracepoint module tainting (kernel 3.15+)
Upstream commit
66cc69e34e86a "Fix: module signature vs tracepoints: add
new TAINT_UNSIGNED_MODULE" fixed an issue where the kernel was
considering unsigned modules as tainting the kernel in the same way as a
force-loaded modules, which was causing the tracepoints within those
modules to be hidden.
This fix was merged in kernel 3.15, so there is no use in applying this
work-around starting from that kernel.
This removes a dependency on kallsyms_lookup_name() for the symbol
"tracepoint_module_notify".
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Tue, 25 Feb 2020 20:12:18 +0000 (15:12 -0500)]
Version 2.12.0-rc2
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Francis Deslauriers [Tue, 4 Feb 2020 21:05:30 +0000 (16:05 -0500)]
Fix: lttng-events.c: variable may be used uninitialized
Fixes the following warning:
/home/frdeso/projets/lttng/modules/lttng-events.c: In function ‘print_metadata_escaped_field’:
/home/frdeso/projets/lttng/modules/lttng-events.c:2563:5: error: ‘ret’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
if (ret)
^
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: If2db5e1ad9782fb49d6d07026976d3d22f89f2ab
Mathieu Desnoyers [Tue, 25 Feb 2020 15:18:11 +0000 (10:18 -0500)]
Fix: rcu: Fix data-race due to atomic_t copy-by-value (5.5.6, 5.4.22)
The following upstream commit has been backported to stable kernels
5.5.6 and 5.4.22:
commit
6cf539a87a61a4fbc43f625267dbcbcf283872ed
Author: Marco Elver <elver@google.com>
Date: Wed Oct 9 17:57:43 2019 +0200
rcu: Fix data-race due to atomic_t copy-by-value
This fixes a data-race where `atomic_t dynticks` is copied by value. The
copy is performed non-atomically, resulting in a data-race if `dynticks`
is updated concurrently.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 11 Feb 2020 20:35:11 +0000 (15:35 -0500)]
fix: workqueue: add worker function to workqueue_execute_end tracepoint (v5.6)
See upstream commit :
commit
1c5da0ec7f20dfb56030fb93f7f52f48e12deb52
Author: Daniel Jordan <daniel.m.jordan@oracle.com>
Date: Mon Jan 13 17:52:39 2020 -0500
workqueue: add worker function to workqueue_execute_end tracepoint
It's surprising that workqueue_execute_end includes only the work when
its counterpart workqueue_execute_start has both the work and the worker
function.
You can't set a tracing filter or trigger based on the function, and
postprocessing scripts interested in specific functions are harder to
write since they have to remember the work from _start and match it up
with the same field in _end.
Add the function name, taking care to use the copy stashed in the
worker since the work is no longer safe to touch.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 11 Feb 2020 20:13:53 +0000 (15:13 -0500)]
fix: media: v4l2: abstract timeval handling in v4l2_buffer (v5.6)
See upstream commit :
commit
77cdffcb0bfb87fe3645894335cb8cb94917e6ac
Author: Arnd Bergmann <arnd@arndb.de>
Date: Mon Dec 16 15:15:00 2019 +0100
media: v4l2: abstract timeval handling in v4l2_buffer
As a preparation for adding 64-bit time_t support in the uapi,
change the drivers to no longer care about the format of the
timestamp field in struct v4l2_buffer.
The v4l2_timeval_to_ns() function is no longer needed in the
kernel after this, but there is userspace code relying on
it to be part of the uapi header.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 11 Feb 2020 20:08:04 +0000 (15:08 -0500)]
fix: rcu: Remove kfree_rcu() special casing and lazy-callback (v5.6)
See upstream commit :
commit
77a40f97030b27b3fc1640a3ed203870f0817f57
Author: Joel Fernandes (Google) <joel@joelfernandes.org>
Date: Fri Aug 30 12:36:32 2019 -0400
rcu: Remove kfree_rcu() special casing and lazy-callback handling
This commit removes kfree_rcu() special-casing and the lazy-callback
handling from Tree RCU. It moves some of this special casing to Tiny RCU,
the removal of which will be the subject of later commits.
This results in a nice negative delta.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 11 Feb 2020 19:51:01 +0000 (14:51 -0500)]
fix: rcu: Fix data-race due to atomic_t copy-by-value (v5.6)
See upstream commit :
commit
6cf539a87a61a4fbc43f625267dbcbcf283872ed
Author: Marco Elver <elver@google.com>
Date: Wed Oct 9 17:57:43 2019 +0200
rcu: Fix data-race due to atomic_t copy-by-value
This fixes a data-race where `atomic_t dynticks` is copied by value. The
copy is performed non-atomically, resulting in a data-race if `dynticks`
is updated concurrently.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 11 Feb 2020 19:47:23 +0000 (14:47 -0500)]
fix: btrfs: make btrfs_ordered_extent naming consistent (v5.6)
See upstream commit :
commit
bffe633e00fb6b904817137fc17a44b42efcd985
Author: Omar Sandoval <osandov@fb.com>
Date: Mon Dec 2 17:34:19 2019 -0800
btrfs: make btrfs_ordered_extent naming consistent with btrfs_file_extent_item
ordered->start, ordered->len, and ordered->disk_len correspond to
fi->disk_bytenr, fi->num_bytes, and fi->disk_num_bytes, respectively.
It's confusing to translate between the two naming schemes. Since a
btrfs_ordered_extent is basically a pending btrfs_file_extent_item,
let's make the former use the naming from the latter.
Note that I didn't touch the names in tracepoints just in case there are
scripts depending on the current naming.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 11 Feb 2020 19:41:29 +0000 (14:41 -0500)]
fix: KVM: x86: Use gpa_t for cr2/gpa to fix TDP support on 32-bit (v5.6)
See upstream commit :
commit
736c291c9f36b07f8889c61764c28edce20e715d
Author: Sean Christopherson <sean.j.christopherson@intel.com>
Date: Fri Dec 6 15:57:14 2019 -0800
KVM: x86: Use gpa_t for cr2/gpa to fix TDP support on 32-bit KVM
Convert a plethora of parameters and variables in the MMU and page fault
flows from type gva_t to gpa_t to properly handle TDP on 32-bit KVM.
Thanks to PSE and PAE paging, 32-bit kernels can access 64-bit physical
addresses. When TDP is enabled, the fault address is a guest physical
address and thus can be a 64-bit value, even when both KVM and its guest
are using 32-bit virtual addressing, e.g. VMX's VMCS.GUEST_PHYSICAL is a
64-bit field, not a natural width field.
Using a gva_t for the fault address means KVM will incorrectly drop the
upper 32-bits of the GPA. Ditto for gva_to_gpa() when it is used to
translate L2 GPAs to L1 GPAs.
Opportunistically rename variables and parameters to better reflect the
dual address modes, e.g. use "cr2_or_gpa" for fault addresses and plain
"addr" instead of "vaddr" when the address may be either a GVA or an L2
GPA. Similarly, use "gpa" in the nonpaging_page_fault() flows to avoid
a confusing "gpa_t gva" declaration; this also sets the stage for a
future patch to combing nonpaging_page_fault() and tdp_page_fault() with
minimal churn.
Sprinkle in a few comments to document flows where an address is known
to be a GVA and thus can be safely truncated to a 32-bit value. Add
WARNs in kvm_handle_page_fault() and FNAME(gva_to_gpa_nested)() to help
document such cases and detect bugs.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 11 Feb 2020 16:20:41 +0000 (11:20 -0500)]
fix: proc: decouple proc from VFS with "struct proc_ops" (v5.6)
See upstream commit :
commit
d56c0d45f0e27f814e87a1676b6bdccccbc252e9
Author: Alexey Dobriyan <adobriyan@gmail.com>
Date: Mon Feb 3 17:37:14 2020 -0800
proc: decouple proc from VFS with "struct proc_ops"
Currently core /proc code uses "struct file_operations" for custom hooks,
however, VFS doesn't directly call them. Every time VFS expands
file_operations hook set, /proc code bloats for no reason.
Introduce "struct proc_ops" which contains only those hooks which /proc
allows to call into (open, release, read, write, ioctl, mmap, poll). It
doesn't contain module pointer as well.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Mon, 24 Feb 2020 20:11:03 +0000 (15:11 -0500)]
fix: y2038: hide timeval/timespec/itimerval/itimerspec types (v5.6)
See upstream commit:
commit
c766d1472c70d25ad475cf56042af1652e792b23
Author: Arnd Bergmann <arnd@arndb.de>
Date: Thu Feb 20 20:03:57 2020 -0800
y2038: hide timeval/timespec/itimerval/itimerspec types
There are no in-kernel users remaining, but there may still be users that
include linux/time.h instead of sys/time.h from user space, so leave the
types available to user space while hiding them from kernel space.
Only the __kernel_old_* versions of these types remain now.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I986a813ad8b1c753ab1fa07f726b0cc481f049cb
Michael Jeanson [Mon, 24 Feb 2020 19:50:20 +0000 (14:50 -0500)]
fix: use timespec64 on kernels that have it
This fixes v5.6 which has hidden 'struct timespec' from kernel code and
makes 32bit archs y2038 compliant on v3.17 and newer.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I1daedc4a93cbbebb8f5a1d99c4619cb27f6a9e07
Michael Jeanson [Mon, 24 Feb 2020 20:15:36 +0000 (15:15 -0500)]
fix: move lttng_close_on_exec to proper wrapper
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I330c6d63fc9c1af00c9a9471563d322d036b9a9c
Michael Jeanson [Mon, 24 Feb 2020 16:30:22 +0000 (11:30 -0500)]
fix: 'struct timex' removed upstream (v5.6)
The 'timex' struct was remove in v5.6 and replaced by 2 variants, one
that is y2038 compliant and a compat version for 32bit archs.
Add this temporary fix while we update our syscalls tracepoint headers,
the type of this struct has limited importance since it's only used to
record the adress in the trace.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Change-Id: I085b22f282db57985f1c3d341e7c0866cb20e3c9
Mathieu Desnoyers [Thu, 20 Feb 2020 15:46:25 +0000 (10:46 -0500)]
Fix: statedump: consistently check task_cred_xxx() return value for NULL
trace_lttng_statedump_process_user_ns() internally checks whether
user_ns is NULL. While this does not appear to be a possible return
value for task_cred_xxx(), err on the safe side and check for NULL here
as well to be consistent with the paranoid behavior of
trace_lttng_statedump_process_user_ns().
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Thu, 20 Feb 2020 14:58:42 +0000 (09:58 -0500)]
Fix: statedump: check task_active_pid_ns return value for NULL
The lttng-statedump checks the return value of task_active_pid_ns()
before each use within lttng_statedump_process_pid_ns(), but misses
the NULL check before dereferencing pid_ns->parent.
This race happens if a task exists in "dead" state while the statedump
iterates on that task.
Reported-by: Li Zhou <li.zhou@windriver.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Geneviève Bastien [Wed, 12 Feb 2020 21:58:25 +0000 (16:58 -0500)]
sched: Make the sched_switch task state an enum
This gives meaning to the task state value. Only the bit masks are
enumerated, as defined compositions are not exhaustive listing of all
possible values and there would be a lot of unknown. Interpretation of
combination of bit flags is left to the consumer of the event.
Change-Id: I83c5fbee9cba2701c7238c0ac6abd4c8a351b193
Signed-off-by: Geneviève Bastien <gbastien@versatic.net>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Geneviève Bastien [Tue, 11 Feb 2020 16:20:27 +0000 (11:20 -0500)]
block: Make the rwbs field as a bit field enum
The rwbs value of block request events is in fact a series of bit fields
set to 0 or 1. The enumeration values are all powers of 2, trace readers
could interpret this as a bit field enum and show the result as a
concatenation of the corresponding flags. For example, with a matching
patch for babeltrace2's pretty sink, the output for a rwbs value of 12
would be:
[13:15:49.
024354958] (+0.
000003868) wilbrod block_rq_complete: { cpu_id = 4 },
{ dev =
8388624, sector =
375490176, nr_sector = 360, error = 0,
rwbs = ( "RWBS_FLAG_READ" | "RWBS_FLAG_RAHEAD" : container = 12 ) }
Consumers who have no notion of the bit field enum can still use the
integer value of the field as before.
Change-Id: I2e873f16e5a0507e52cfc85c70b471a9f4d5d5d1
Signed-off-by: Geneviève Bastien <gbastien@versatic.net>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 5 Feb 2020 15:36:41 +0000 (10:36 -0500)]
Version 2.12.0-rc1
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Francis Deslauriers [Tue, 28 Jan 2020 22:19:20 +0000 (17:19 -0500)]
Fix: lttng-syscalls.c: marking wrong syscall probe as unregistered
When calling `lttng_syscalls_unregister()` we currently mark as
unregistered the wrong syscall probe type.
Concretely, when we unregister "sys_exit" we wrongfully mark that
sys_enter is unregistered and vice versa.
Given than currently entry and exit probes are always enabled together
(except on internal errors), the effect of this bug is not user-visible.
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Change-Id: I64acf6a941a3d1fa1bf8be424f834ddb7fb92ace
Mathieu Desnoyers [Tue, 7 Jan 2020 18:09:38 +0000 (13:09 -0500)]
Version 2.12.0-pre
Integrate the missing 2.11.0-rc1 entries to the changelog. This
was caused by forking to stable-2.11 the commit before the v2.11.0-rc1
tag. In the future, we will try to keep the rc1 tag as the common
ancestor between master and stable branches.
This also updates the lttng-modules version to 2.12.0-pre, which is
a state that will stay until the v2.12.0-rc1 tag.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Jérémie Galarneau [Fri, 20 Dec 2019 03:11:51 +0000 (22:11 -0500)]
Bump LTTNG_MODULES_ABI_MINOR_VERSION to 5
New operations were added to the lttng-modules ABI as part of the 2.12
release cycle to support UID tracking and the "session clear"
functionality.
This will allow future LTTng-tools versions to check for those
capabilities.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 17 Dec 2019 17:11:11 +0000 (12:11 -0500)]
fix: use user ns wrapper code in new id trackers
These wrappers are required to translate kuid on kernels prior to v3.5.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Michael Jeanson [Tue, 17 Dec 2019 17:11:10 +0000 (12:11 -0500)]
fix: function prototype in wrapper/mm.h
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Tue, 17 Dec 2019 15:17:01 +0000 (10:17 -0500)]
ID tracker: implement vpid/uid/vuid/gid/vgid trackers
Co-developed-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
This page took 0.052723 seconds and 4 git commands to generate.