Lukas Bulwahn [Thu, 29 Feb 2024 03:07:42 +0000 (04:07 +0100)]
docs: submit-checklist: use subheadings
During review (see Link), Jani Nikula suggested to use proper subheadings
instead of using italics to indicate the different new top-level
categories in the checklist. Further the top heading should follow the
common scheme.
Use subheadings. Adjust to common heading adornment.
Lukas Bulwahn [Thu, 29 Feb 2024 03:07:41 +0000 (04:07 +0100)]
docs: submit-checklist: structure by category
While going through the submit checklist, the list order seemed rather
random, probably just by historical coincidences of always adding yet the
next point someone thought of at the end of the list.
Structure and order them by the category of such activity,
reviewing, documenting, checking with tools, building and testing.
As the diff of the reordering is large:
Review code now includes previous points 1, 5 and 22.
Review Kconfig includes previous 6, 7 and 8.
Documenting includes previous 11, 15, 16, 17, 18 and 23.
Checking with tools includes previous 5, 9 and 10.
Building includes previous 2, 3, 20 and 24.
Testing includes previous 12, 13, 14, 19 and 21.
Previous point 4 (compile for ppc64) was merged into point 3 (build for
many architectures), as it was just a further note to cross-compiling.
Previous point 5 was split into one in review and one in checking
to have every previous point in the right category.
Point 11 was shortened, as building documentation is mentioned already
in Build your code, 1d.
A note that was presented visually much too aggressive in the HTML view was
turned into a simple "Note that..." sentence in the enumeration.
The recommendation to test with the -mm patchset (previous 21, now
testing, point 5) was updated to the current state of affairs to test with
a recent tag of linux-next.
Note that the previous first point still remains the first list even after
reordering. Randy confirmed that it was important to Stephen Rothwell to
keep 'include what you use' to be the first in the list.
While at it, replace the reference to the obsolete CONFIG_DEBUG_SLAB with
CONFIG_SLUB_DEBUG.
Reviewed-by: Randy Dunlap <rdunlap@infradead.org> Tested-by: Randy Dunlap <rdunlap@infradead.org> Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com> Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Message-ID: <20240229030743.9125-2-lukas.bulwahn@gmail.com>
Thorsten Leemhuis [Fri, 1 Mar 2024 08:41:06 +0000 (09:41 +0100)]
docs: new text on bisecting which also covers bug validation
Add a second document on bisecting regressions explaining the whole
process from beginning to end -- while also describing how to validate
if a problem is still present in mainline. This "two in one" approach
is possible, as checking whenever a bug is in mainline is one of the
first steps before performing a bisection anyway and thus needs to be
described. Due to this approach the text also works quite nicely in
conjunction with Documentation/admin-guide/reporting-issues.rst, as it
covers all typical cases where users will need to build a kernel in
exactly the same order.
The text targets users that normally run kernels from their Linux
distributor who might never have compiled their own kernel.
This aim is why the first kernel built while following this guide is
generated from the latest mainline codebase. This will rule out that the
regression (a) was fixed already and (b) is caused by config change a
vendor distributor performed; checking mainline will furthermore (c)
determine if the issue is something that needs to be reported to the
regular developers or the stable team (this is needed even when readers
bisect within a stable series).
Only then are readers instructed to build their own variant of the
'good' kernel to validate the trimmed .config file created during early
in the guide, as performing a bisection with a broken one would be a
waste of time. There is a small downside of this order: readers might
have to go back to testing mainline, if it turns out there is a problem
with their .config. But that should be rare -- and if the regression was
already fixed readers might not get to this point anyway. Hence in the
end this order should mean that readers built less kernels overall.
This sequence allows the text to easily cover the "check if a bug is
present in the upstream kernel" case while only making things a tiny bit
more complicated.
The text tries to prevent readers from running into many mistakes users
are known to frequently make. The steps required for this might look
superfluous for people that are already familiar with bisections -- but
anyone with that knowledge should be able to adapt the instructions to
their use-case or will not need this text at all.
Style and structure of the text match the one
Documentation/admin-guide/quickly-build-trimmed-linux.rst uses. Quite a
few paragraphs are even copied from there and not changed at all or only
slightly. This will complicate maintenance, as some future changes to
one of these documents will have to be replicated in the other. But this
is the lesser evil: solutions like "sending readers from one document
over to the other" or "extracting the common parts into a separate
document" might work in other cases, but would be too confusing here
given the topic and the target audience.
Signed-off-by: Thorsten Leemhuis <linux@leemhuis.info>
[jc: Undo spurious removal of subsection header line] Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Message-ID: <02b084a06de4ad61ac4ecd92b9265d4df4d03d71.1709282441.git.linux@leemhuis.info>
Lukas Bulwahn [Fri, 1 Mar 2024 14:18:00 +0000 (15:18 +0100)]
docs: drop the version constraints for sphinx and dependencies
As discussed (see Links), there is some inertia to move to the recent
Sphinx versions for the doc build environment.
As first step, drop the version constraints and the related comments. As
sphinx depends on jinja2, jinja2 is pulled in automatically. So drop that.
Then, the sphinx-pre-install script will fail though with:
Can't get default sphinx version from ./Documentation/sphinx/requirements.txt at ./scripts/sphinx-pre-install line 305.
The script simply expects to parse a version constraint with Sphinx in the
requirements.txt. That version is used in the script for suggesting the
virtualenv directory name.
To suggest a virtualenv directory name, when there is no version given in
the requirements.txt, one could try to guess the version that would be
downloaded with 'pip install -r Documentation/sphinx/requirements.txt'.
However, there seems no simple way to get that version without actually
setting up the venv and running pip. So, instead, name the directory with
the fixed name 'sphinx_latest'.
Finally update the Sphinx build documentation to reflect this directory
name change.
Akira Yokosawa [Sun, 25 Feb 2024 09:46:00 +0000 (18:46 +0900)]
docs: Restore "smart quotes" for quotes
Commit eaae75754d81 ("docs: turn off "smart quotes" in the HTML build")
disabled conversion of quote marks along with that of dashes.
Despite the short summary, the change affects not only HTML build
but also other build targets including PDF.
However, as "smart quotes" had been enabled for more than half a
decade already, quite a few readers of HTML pages are likely expecting
conversions of "foo" -> “foo” and 'bar' -> ‘bar’.
Furthermore, in LaTeX typesetting convention, it is common to use
distinct marks for opening and closing quote marks.
To satisfy such readers' expectation, restore conversion of quotes
only by setting smartquotes_action [1].
Carlos Bilbao [Tue, 9 Jan 2024 15:56:43 +0000 (09:56 -0600)]
docs: Include simplified link titles in main index
Include simplified link titles in the main page's documentation index to
enhance website's readability and UX. Update the text that directs users to
various documents without changing the actual titles chosen by the authors.
Carlos Bilbao [Tue, 9 Jan 2024 15:56:42 +0000 (09:56 -0600)]
docs: Correct formatting of title in admin-guide/index.rst
Adjust the title of "The Linux kernel user's and administrator's guide" to
adhere to the expected reStructuredText (rst) formatting, using double
equal signs for the main header.
Vegard Nossum [Mon, 5 Feb 2024 17:51:26 +0000 (18:51 +0100)]
docs: kernel_feat.py: fix build error for missing files
If the directory passed to the '.. kernel-feat::' directive does not
exist or the get_feat.pl script does not find any files to extract
features from, Sphinx will report the following error:
Sphinx parallel build error:
UnboundLocalError: local variable 'fname' referenced before assignment
make[2]: *** [Documentation/Makefile:102: htmldocs] Error 2
This is due to how I changed the script in c48a7c44a1d0 ("docs:
kernel_feat.py: fix potential command injection"). Before that, the
filename passed along to self.nestedParse() in this case was weirdly
just the whole get_feat.pl invocation.
We can fix it by doing what kernel_abi.py does -- just pass
self.arguments[0] as 'fname'.
SeongJae Park [Fri, 16 Feb 2024 20:19:02 +0000 (12:19 -0800)]
MAINTAINERS: Set the field name for subsystem profile section
Subsystem profile section entry identifier is not having its field name
that can be parsed by maintainers_include.py, unlike other sections
which have their own human-readable field names. As a result, profile
sections on rendered rst file is having weird name, 'P:'. Set the field
name as 'Subsystem Profile'.
Ran.Park [Wed, 14 Feb 2024 11:40:08 +0000 (19:40 +0800)]
Fixed case issue with 'fault-injection' in documentation
In the 'fault-injection' subdirectory, the first letter F
is capitalized, whereas in index.rst f is lowercase, but in
index.rst all other elements in the same column are capitalized.
Johannes Berg [Wed, 14 Feb 2024 13:29:38 +0000 (14:29 +0100)]
kernel-doc: handle #if in enums as well
In addition to #ifdef, #define and #endif, also handle
any #if since we may be using e.g. #if IS_ENABLED(...).
I didn't find any instances of this in the kernel now,
there are enums with such ifs inside, but I didn't find
any with kernel-doc as well. However, it came up as we
were adding such a construct in our driver and warnings
from kernel-doc were the result.
Konstantin Ryabitsev [Wed, 14 Feb 2024 20:09:53 +0000 (15:09 -0500)]
Documentation: update mailing list addresses
The mailman2 server running on lists.linuxfoundation.org will be shut
down in very imminent future. Update all instances of obsolete list
addresses throughout the tree with their new destinations.
Vegard Nossum [Thu, 15 Feb 2024 13:48:26 +0000 (14:48 +0100)]
scripts/kernel-doc: separate out function signature
Format the entire function signature and place it in a separate variable;
this both makes it easier to understand what these lines of code are doing
and will allow us to simplify the code further in the following patch.
Vegard Nossum [Thu, 15 Feb 2024 06:41:09 +0000 (07:41 +0100)]
docs: translations: use attribute to store current language
Akira Yokosawa reported [1] that the "translations" extension we added in
commit 7418ec5b151f ("docs: translations: add translations links when they
exist") broke the build on Sphinx versions v6.1.3 through 7.1.2 (possibly
others) with the following error:
Exception occurred:
File "/usr/lib/python3.12/site-packages/sphinx/util/nodes.py", line 624, in _copy_except__document
newnode = self.__class__(rawsource=self.rawsource, **self.attributes)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: LanguagesNode.__init__() missing 1 required positional argument: 'current_language'
The full traceback has been saved in /tmp/sphinx-err-7xmwytuu.log, if you want to report the issue to the developers.
Solve this problem by making 'current_language' a true element attribute
of the LanguagesNode element, which is probably the more correct way to do
it anyway.
Jonathan Corbet [Mon, 19 Feb 2024 16:05:38 +0000 (09:05 -0700)]
docs: Instruct LaTeX to cope with deeper nesting
The addition of the XFS online fsck documentation starting with
commit a8f6c2e54ddc ("xfs: document the motivation for online fsck design")
added a deeper level of nesting than LaTeX is prepared to deal with. That
caused a pdfdocs build failure with the helpful "Too deeply nested" error
message buried deeply in Documentation/output/filesystems.log.
Increase the "maxlistdepth" parameter to instruct LaTeX that it needs to
deal with the deeper nesting whether it wants to or not.
Hunter Chasens [Wed, 7 Feb 2024 17:10:07 +0000 (12:10 -0500)]
docs: admin-guide: Update bootloader and installation instructions
Updates the bootloader and installation instructions in
admin-guide/README.rst to align with modern practices.
Details of Changes:
- Added guidance on using EFISTUB for UEFI/EFI systems.
- Noted that LILO is no longer in active development and provides
alternatives.
- Kept LILO instructions but marked as Legacy LILO Instructions.
Suggest removal in future patch.
Thorsten Blum [Thu, 8 Feb 2024 14:50:01 +0000 (15:50 +0100)]
docs: Makefile: Fix make cleandocs by deleting generated .rst files
The script tools/net/ynl/ynl-gen-rst.py (YNL_TOOL) generates several .rst
files (YNL_INDEX, YNL_RST_FILES) in Documentation/networking/netlink_spec
(YNL_RST_DIR) which are not deleted by make cleandocs.
Fix make cleandocs by deleting the generated .rst files.
Vegard Nossum [Thu, 8 Feb 2024 16:17:05 +0000 (17:17 +0100)]
scripts/kernel-doc: reindent
This file is using an ungodly mixture of 4 spaces, 2-wide tabs, 4-wide
tabs, _and_ 8-wide tabs, making it really hard to find good editor
settings for working with this file.
Bite the bullet and reindent it by hand. I tried using both perltidy
and vim, but neither of them were up to the task without changing too
much or getting confused about what they were supposed to be doing.
I did change a few instances of
}
else
into
} else
(and same for elsif); the file is again written using both styles, and
I left functions which already seemed self-consistent alone.
You can verify that this commit only changes whitespace using e.g.:
git diff --ignore-all-space --word-diff
or to see (only) the instances where newlines were added/removed:
git diff --ignore-all-space
You can also see the delta from what perltidy would have wanted to
do to this file (when asked to only indent it), which isn't that much
in the end:
Thorsten Blum [Thu, 8 Feb 2024 20:55:51 +0000 (21:55 +0100)]
docs: scripts: sphinx-pre-install: Fix building docs with pyyaml package
The Python module pyyaml is required to build the docs, but it is only
listed in Documentation/sphinx/requirements.txt and is therefore missing
when Sphinx is installed as a package and not via pip/pypi.
Add pyyaml as an optional package for multiple distros to fix building the
docs if you prefer to install Sphinx as a package.
Federico Vaga [Fri, 9 Feb 2024 22:21:15 +0000 (23:21 +0100)]
doc:it_IT: remove unreferenced and not translated page
The netdev-FAQ page in the italian translation was creted to avoid
having broken links. With the evolution of the documentation this was
not referenced anymore, but the page never removed.
The servers for the @codeaurora domain have long been retired and any
messages addressed to @codeaurora will bounce.
Trilok has an entry in .mailmap, but the raw documentation files still
list an old @codeaurora address. Update the address in the
documentation files for anyone reading them.
Vegard Nossum [Mon, 5 Feb 2024 17:51:26 +0000 (18:51 +0100)]
docs: kernel_feat.py: fix build error for missing files
If the directory passed to the '.. kernel-feat::' directive does not
exist or the get_feat.pl script does not find any files to extract
features from, Sphinx will report the following error:
Sphinx parallel build error:
UnboundLocalError: local variable 'fname' referenced before assignment
make[2]: *** [Documentation/Makefile:102: htmldocs] Error 2
This is due to how I changed the script in c48a7c44a1d0 ("docs:
kernel_feat.py: fix potential command injection"). Before that, the
filename passed along to self.nestedParse() in this case was weirdly
just the whole get_feat.pl invocation.
We can fix it by doing what kernel_abi.py does -- just pass
self.arguments[0] as 'fname'.
Guilherme G. Piccoli [Sat, 3 Feb 2024 15:21:13 +0000 (12:21 -0300)]
docs: Document possible_cpus parameter
The number of possible CPUs is set be kernel in early boot time through
some discovery mechanisms, like ACPI in x86. We have a parameter both
in x86 and S390 to override that - there are some cases of BIOSes exposing
more possible CPUs than the available ones, so this parameter is a good
testing mechanism, but for some reason wasn't mentioned so far in the
kernel parameters guide - let's fix that.
Sakari Ailus [Wed, 31 Jan 2024 08:49:34 +0000 (10:49 +0200)]
kernel-doc: Support arrays of pointers struct fields
In a rather unusual arrangement in include/media/v4l2-vp9.h struct
v4l2_vp9_frame_symbol_counts has fields that are arrays of pointers, not a
pointer to an array, which is what's usually done.
Add support for such arrays of pointers to kernel-doc.
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com> Acked-by: Randy Dunlap <rdunlap@infradead.org> Tested-by: Randy Dunlap <rdunlap@infradead.org> Tested-by: Ricardo Ribalda <ribalda@chromium.org> Signed-off-by: Jonathan Corbet <corbet@lwn.net> Link: https://lore.kernel.org/r/20240131084934.191226-1-sakari.ailus@linux.intel.com
Abhishek Pandit-Subedi [Fri, 26 Jan 2024 00:53:11 +0000 (16:53 -0800)]
coding-style: Add guidance to prefer dev_dbg
During review, it was suggested that drivers only emit messages when
something is wrong or it is a debug message. Document this as a formal
recommendation.
Michael Kelley [Thu, 25 Jan 2024 04:59:41 +0000 (20:59 -0800)]
docs: Fix subsystem APIs page so ungrouped entries have their own header
The kernel subsystem APIs front page currently has four top-level
groupings with headers, and then everything else that isn't grouped.
But in the table-of-contents, ungrouped subsystems are indented as
if they were part of the preceding grouping (currently "Storage
interfaces"), which is confusing.
Fix this by adding an "Other subsystems" header for the ungrouped
subsystems.
Andrew Ballance [Wed, 24 Jan 2024 04:39:18 +0000 (22:39 -0600)]
docs: sphinx-pre-install fix-noto-sans-cjk on fedora
fedora 38 and later changed the directory and package name that
provides NotoSansCJK-Regular.ttc. this adds the new search path and
suggests the correct package if on fedora 38 or later.
Leo Yan [Tue, 2 Jan 2024 08:50:01 +0000 (16:50 +0800)]
Documentation: userspace-api: Document perf ring buffer mechanism
In the Linux perf tool, the ring buffer serves not only as a medium for
transferring PMU event data but also as a vital mechanism for hardware
tracing using technologies like Intel PT and Arm CoreSight, etc.
Consequently, the ring buffer mechanism plays a crucial role by ensuring
high throughput for data transfer between the kernel and user space
while avoiding excessive overhead caused by the ring buffer itself.
This commit documents the ring buffer mechanism in detail. It explains
the implementation of both the regular ring buffer and the AUX ring
buffer. Additionally, it covers how these ring buffers support various
tracing modes and explains the synchronization with memory barriers.
Jonathan Corbet [Mon, 22 Jan 2024 22:18:30 +0000 (15:18 -0700)]
docs: rework the userspace-api top page
Add some subsection headings and reorder entries so that the page makes a
bit more sense. With luck, adding some ordering will also reduce merge
conflicts due to everybody adding new entries at the end.
Jonathan Corbet [Mon, 22 Jan 2024 22:19:19 +0000 (15:19 -0700)]
docs: rework the driver-api top-level page
Add subsections in an attempt to bring a bit order to this page; also sort
most subsections into alphabetical order. With luck all this will help to
prevent merge conflicts on this page due to everybody adding entries at the
end.
Randy Dunlap [Sun, 7 Jan 2024 01:24:00 +0000 (17:24 -0800)]
doc-guide: kernel-doc: tell about object-like macros
Since 2014 kernel-doc has supported describing object-like macros
but it is not documented anywhere. I should have required some
documentation for it when I merged the patch. :(
There are currently only 3 uses of this (all in DRM headers, in
include/drm/*.h).
Add object-like macro kernel-doc documentation now so that more may
know about it and use it.
Fixes: cbb4d3e6510b ("scripts/kernel-doc: handle object-like macros") Signed-off-by: Randy Dunlap <rdunlap@infradead.org> Cc: Jonathan Corbet <corbet@lwn.net> Cc: linux-doc@vger.kernel.org Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Jonathan Corbet <corbet@lwn.net> Link: https://lore.kernel.org/r/20240107012400.32587-1-rdunlap@infradead.org
Vegard Nossum [Thu, 11 Jan 2024 09:48:38 +0000 (10:48 +0100)]
docs: add blurb about target audience to maintainer-profile
It's good to be clear about who the intended target audience for any
given piece of documentation is, as this will help us put new text in
the correct place. Let's encourage submitters to state it explicitly
rather than relying on where they placed it in the directory hierarchy
as there isn't necessarily a one-to-one correspondence between them.
Target audience: documentation contributors and reviewers.
Reformat lines in kernel-doc comments, which make use of the backslash at
the end to suggest it is a multi-line comment. kernel-doc is able to
process e.g. the short description of a function properly, even if it is
across two lines.
Lukas Bulwahn [Thu, 30 Nov 2023 09:55:15 +0000 (10:55 +0100)]
docs: admin-guide: remove obsolete advice related to SLAB allocator
Commit 1db9d06aaa55 ("mm/slab: remove CONFIG_SLAB from all Kconfig and
Makefile") removes the config SLAB and makes the SLUB allocator the only
default allocator in the kernel. Hence, the advice on reducing OS jitter
due to kworker kernel threads to build with CONFIG_SLUB instead of
CONFIG_SLAB is obsolete.
Remove the obsolete advice to build with SLUB instead of SLAB.
Gustavo Sousa [Tue, 23 Jan 2024 16:21:58 +0000 (13:21 -0300)]
docs/sphinx: Fix TOC scroll hack for the home page
When on the documentation home page, there won't be any ".current"
element since no entry from the TOC was selected yet. That results in a
javascript error. Fix that by only trying to set the scrollTop if we
have matches for current entries.
Linus Torvalds [Sun, 21 Jan 2024 22:01:12 +0000 (14:01 -0800)]
Merge tag 'bcachefs-2024-01-21' of https://evilpiepirate.org/git/bcachefs
Pull more bcachefs updates from Kent Overstreet:
"Some fixes, Some refactoring, some minor features:
- Assorted prep work for disk space accounting rewrite
- BTREE_TRIGGER_ATOMIC: after combining our trigger callbacks, this
makes our trigger context more explicit
- A few fixes to avoid excessive transaction restarts on
multithreaded workloads: fstests (in addition to ktest tests) are
now checking slowpath counters, and that's shaking out a few bugs
- Assorted tracepoint improvements
- Starting to break up bcachefs_format.h and move on disk types so
they're with the code they belong to; this will make room to start
documenting the on disk format better.
- A few minor fixes"
* tag 'bcachefs-2024-01-21' of https://evilpiepirate.org/git/bcachefs: (46 commits)
bcachefs: Improve inode_to_text()
bcachefs: logged_ops_format.h
bcachefs: reflink_format.h
bcachefs; extents_format.h
bcachefs: ec_format.h
bcachefs: subvolume_format.h
bcachefs: snapshot_format.h
bcachefs: alloc_background_format.h
bcachefs: xattr_format.h
bcachefs: dirent_format.h
bcachefs: inode_format.h
bcachefs; quota_format.h
bcachefs: sb-counters_format.h
bcachefs: counters.c -> sb-counters.c
bcachefs: comment bch_subvolume
bcachefs: bch_snapshot::btime
bcachefs: add missing __GFP_NOWARN
bcachefs: opts->compression can now also be applied in the background
bcachefs: Prep work for variable size btree node buffers
bcachefs: grab s_umount only if snapshotting
...
Linus Torvalds [Sun, 21 Jan 2024 19:14:40 +0000 (11:14 -0800)]
Merge tag 'timers-core-2024-01-21' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull timer updates from Thomas Gleixner:
"Updates for time and clocksources:
- A fix for the idle and iowait time accounting vs CPU hotplug.
The time is reset on CPU hotplug which makes the accumulated
systemwide time jump backwards.
- Assorted fixes and improvements for clocksource/event drivers"
* tag 'timers-core-2024-01-21' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
tick-sched: Fix idle and iowait sleeptime accounting vs CPU hotplug
clocksource/drivers/ep93xx: Fix error handling during probe
clocksource/drivers/cadence-ttc: Fix some kernel-doc warnings
clocksource/drivers/timer-ti-dm: Fix make W=n kerneldoc warnings
clocksource/timer-riscv: Add riscv_clock_shutdown callback
dt-bindings: timer: Add StarFive JH8100 clint
dt-bindings: timer: thead,c900-aclint-mtimer: separate mtime and mtimecmp regs
Kent Overstreet [Tue, 16 Jan 2024 21:20:21 +0000 (16:20 -0500)]
bcachefs: opts->compression can now also be applied in the background
The "apply this compression method in the background" paths now use the
compression option if background_compression is not set; this means that
setting or changing the compression option will cause existing data to
be compressed accordingly in the background.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Kent Overstreet [Tue, 16 Jan 2024 18:29:59 +0000 (13:29 -0500)]
bcachefs: Prep work for variable size btree node buffers
bcachefs btree nodes are big - typically 256k - and btree roots are
pinned in memory. As we're now up to 18 btrees, we now have significant
memory overhead in mostly empty btree roots.
And in the future we're going to start enforcing that certain btree node
boundaries exist, to solve lock contention issues - analagous to XFS's
AGIs.
Thus, we need to start allocating smaller btree node buffers when we
can. This patch changes code that refers to the filesystem constant
c->opts.btree_node_size to refer to the btree node buffer size -
btree_buf_bytes() - where appropriate.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
In __bch2_ioctl_subvolume_create(), we grab s_umount unconditionally
and unlock it at the end of the function. There is a comment
"why do we need this lock?" about the lock coming from
commit 42d237320e98 ("bcachefs: Snapshot creation, deletion")
The reason is that __bch2_ioctl_subvolume_create() calls
sync_inodes_sb() which enforce locked s_umount to writeback all dirty
nodes before doing snapshot works.
Fix it by read locking s_umount for snapshotting only and unlocking
s_umount after sync_inodes_sb().
Signed-off-by: Su Yue <glass.su@suse.com> Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Colin Ian King [Tue, 16 Jan 2024 11:07:23 +0000 (11:07 +0000)]
bcachefs: remove redundant variable tmp
The variable tmp is being assigned a value but it isn't being
read afterwards. The assignment is redundant and so tmp can be
removed.
Cleans up clang scan build warning:
warning: Although the value stored to 'ret' is used in the enclosing
expression, the value is never actually read from 'ret'
[deadcode.DeadStores]
Signed-off-by: Colin Ian King <colin.i.king@gmail.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Kent Overstreet [Tue, 16 Jan 2024 01:37:23 +0000 (20:37 -0500)]
bcachefs: Fix excess transaction restarts in __bchfs_fallocate()
drop_locks_do() should not be used in a fastpath without first trying
the do in nonblocking mode - the unlock and relock will cause excessive
transaction restarts and potentially livelocking with other threads that
are contending for the same locks.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Kent Overstreet [Mon, 15 Jan 2024 22:59:51 +0000 (17:59 -0500)]
bcachefs: Better journal tracepoints
Factor out bch2_journal_bufs_to_text(), and use it in the
journal_entry_full() tracepoint; when we can't get a journal reservation
we need to know the outstanding journal entry sizes to know if the
problem is due to excessive flushing.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Kent Overstreet [Mon, 15 Jan 2024 22:56:22 +0000 (17:56 -0500)]
bcachefs: Avoid flushing the journal in the discard path
When issuing discards, we may need to flush the journal if there's too
many buckets that can't be discarded until a journal flush.
But the heuristic was bad; we should be comparing the number of buckets
that need to flushes against the number of free buckets, not the number
of buckets we saw.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Kent Overstreet [Mon, 15 Jan 2024 19:15:26 +0000 (14:15 -0500)]
bcachefs: bch2_kthread_io_clock_wait() no longer sleeps until full amount
Drop t he loop in bch2_kthread_io_clock_wait(): this allows the code
that uses it to be woken up for other reasons, and fixes a bug where
rebalance wouldn't wake up when a scan was requested.
This raises the possibility of spurious wakeups, but callers should
always be able to handle that reasonably well.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
projects / users / dwmw2 / linux.git / log