Previously, lower was better, and higher was worse. But really we want
the scores to be based primarily around the index sizes, which can
grow to be very large, so a maximum score is hard to determine.
Instead, start with the index size, and then divide to make it smaller
based on how "useful" the equation terms will be in searching it.
Improves the performance of queries like those in #19080; according
to humdinger's testing, the query with the most expensive term first
went from ~2.0s execution time down to ~0.7s, same as the query with
the least expensive term first.
Change-Id: Id71fa21c95cfe3d8d0019ff356bdf4935446411f
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8593
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
This amends 0f42cf8d7d6904e9c39c6cda655f0e3dd90a6d91.
That commit accidentally contained an unrelated patch to the one
described in the commit message, turning a number of prints that
happened unconditionally to ones that happened only for faults
occurring in kernel context. This change re-enables those prints
when TRACE_FAULTS is set (which it isn't by default.)
The original change to reduce prints was done to cut down on syslog spam
from applications like Boehm-GC, the JVM, or emulators that trigger
many (thousands or more) faults on purpose and then catch them in
signal handlers. In the case where these faults are crashes, then
the debug_server will still log those crashes into the syslog anyway.
And handle another corner case in set_memory_protection.
This should fix the remaining causes of #19012.
Change-Id: I92fc184024ffb3923239da5e0bd90efc3085da28
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8589
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
B_PATH_NAME_LENGTH == PATH_MAX, and PATH_MAX is inclusive of the final
NULL terminator, so we don't need a + 1 here.
The original KPath default was to not use + 1, but that was changed in
42e3c6f97874f37701385e7027c77e4366d7c450 due to all the consumers that did.
But all those consumers are wrong, it appears; they should just be
using the default length instead. So now we do that.
* If we coerce types inside the switch(), then the "type already converted"
check at the beginning will fail every time, causing us to reconvert,
which is surely bad for performance.
* B_TIME_TYPE should be INT32 or INT64 depending on what its size is.
May help with #19080.
Its values must be shifted before comparing against them.
Also handle the last-modified times correctly in NodeGetLastModifiedTime.
Fixes an issue noticed in #19080 which was a regression from
the query refactor earlier this year.
Also while at it, remove a needless lock in EntryGetName;
Inode::GetName acquires this lock for us.
We pass a NULL destination address, so we need to keep the added
NULL check, but after the previous commit we can now use the
real hook function here.
WiFi still works (tested with realtekwifi.)
This is what FreeBSD does. The default implementation (in this
same file) does what ether_output did before, but the FreeBSD
net80211 stack overrides the method.
IORequest notifications begin by notifying the "finished" condition,
then invoking the request callback, then invoking the parent callback.
vfs_{read|write}_pages wait on the "finished" condition and then
return at once, potentially releasing their references to the vnode
in question, before the request callback is even invoked. The request
callback is what (eventually) invokes iterative_io_finished_hook,
which meant that we were accessing the Inode object after our reference
to it had already been released.
We can't really change IORequest notification order, as any one of the
notifications could delete the request, and indeed the first one does here.
So the solution is just to acquire another reference to the vnode and
release it in the finished hook.
Fixes #19122 and #8405.
* Use do_fd_io and drop the unnecessary iteration callback.
* Make the callback call a non-static function to avoid needing
to fetch values.
* Unlock the cache before calling the I/O function.
Should fix #19259.
If get_vecs returns 0 and no vecs, then we need to quit iterating.
Otherwise we'll just loop forever and leak memory.
Also add an assert in IOBuffer::GetNextVirtualVec that would have caught
this problem, and the memory leak it results in.
We need to hold a reference to the file descriptor throughout
the I/O operation, which do_iterative_fd_io already takes care of.
So just modify it to handle not having the additional callbacks.
Nothing actually uses do_fd_io at present, though it will be
in the next commits.
* Spirograph uses SetBits with length 1 und 4, which can't work as
expected. As Spirograph uses the same source and target colorspace,
the memcpy codepath is used with the right length.
* fix #12931
Change-Id: I4f27e748fb842c2d81d4161e7c94586844995bdf
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8562
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: Adrien Destugues <pulkomandy@pulkomandy.tk>
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Previously it was not initialized until "post-VM", but there are
a number of ways VM initialization can go wrong that it would
be nice to know about without needing a serial port.
On arches which map the whole physical memory into the kernel
address space (x86_64, at least), we can get the bluescreen facility
initialized using KERNEL_PMAP_BASE. On other architectures, we
just fail to init then, and do the usual setup later on.
A slight bit of extra code cleanup in blue_screen_init_early:
we now just call module->info.std_ops() rather than a
frame-buffer-console specific method.
If the vnode's reference count is already > 0 and we can increment
it from there, then we know it's already used and we don't need
to run any further checks. This saves us having to acquire the
vnode lock every single time we want to get the vnode.
(IsBusy() doesn't use an atomic get to fetch the flags value,
but as this is the first read of the value after having acquired
the vnode lock, that should not be a problem.)
Adjust free_vnode, while at it, to not set a positive reference count
during destruction, to be fully certain nothing will see this vnode
as being available for use, and add an assertion checking this.
Improves "git status" time significantly, at least in the case where
all vnodes are in the disk caches. This fast path also seems to be hit
very many times (> 100,000) on boot, and in compile jobs, too.
Change-Id: Ibcc65aecbfcdc4f3fca42baa3756e39aab8b9efb
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8583
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
When the transfer is canceled, there is no new data received. So we
should stop the processing and inform the network stack that the
transfer is canceled.
Change-Id: I5c991899292b84e6d9bc589348a5ab32384ea09a
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8577
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
The MAC address is randomly generated each time the device is connected.
So there is no way to match with a previous session and reattach the
interface transparently.
Change-Id: I8aea95d5a09621a0dcdd7ce89787663a38435001
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8576
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: Adrien Destugues <pulkomandy@pulkomandy.tk>
This lock protects a linked-list. In all cases but one, the only
operation done while holding the lock is to remove a single item
from the list and decrement a counter. Acquiring a mutex itself
involves multiple linked-list operations protected by spinlocks,
so cut out the overhead and just use a spinlock directly.
In the one case where we do more than just remove an item, we
hold an additional write-lock, and so we don't run any risk of
causing "spinlock could not be acquired for a long time" KDLs,
as in that case the threads will be waiting on the rwlock instead.
Reduces lock contention in the VFS. Compiling HaikuDepot and the
mime_db with -j4 (in a VM), the sys time decreased a bit (~10.1s
to ~9.9s), and real time went down by more (~31s to ~29s.) "git status"
performance also improved a bit, but we seem to be contending for
vnode locks now in that case.
Avoids needing a nonstandard "offsetof", and inlines more methods.
No behavioral change intended.
Also rename "next" to "hash_next" for clarity, and remove an unused
"next" field from LockWaiter.
Applications that don't call open() or like functions too often,
and call many FD-related methods across multiple threads at once
(like "git status") now don't wait on the context lock as much.
("git status" performance isn't much improved because threads just
hit the "unused vnodes" lock instead.)
- Convert libroot from a page to a group, so that it can list the
included file and functions in the generated book (like the kits)
- Add unistd.dox and move the relevant part of ioctl details there
- Make sure to use C89-compatible comments only in POSIX headers
Change-Id: I8f0412e4c75de6f48018a0436909f8b0076342a4
Reviewed-on: https://review.haiku-os.org/c/haiku/+/6369
Reviewed-by: Adrien Destugues <pulkomandy@pulkomandy.tk>
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
Exposes a new field on mouse events named "device" that specifices what
type of device the event originates from. This is then used in
PadBlocker to ignore events that don't originate from trackpads.
This field should also be exposed to the public API in case any program
in userspace want to use it, as it may be useful for other purposes.
Fixes #19238 and #17821
Change-Id: Ic0f8c088cf5d2b0414a8e69498b48f2a3e370d73
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8557
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: Adrien Destugues <pulkomandy@pulkomandy.tk>
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
The driver had its own header file with definitions from the USB video
specification. Use the system ones already in place for use in listusb.
Also recognize devices which advertise themselves as "miscellaneous" in
the device descriptor, this is the recommended way in the specification
as the usb_video interface may be only part of a device, with other
interfaces and endpoints used for audio or for HID for example.
Change-Id: I7e2e45328dcc1e81c407937e8dd3d77209c5c52a
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8581
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Superblock field blocks_per_ag is confusing, as it holds the number
of bitmap blocks that are in each allocation group, not the number
of disk blocks per allocation group. Try to clarify when possible.
Change-Id: I60dad9d3c5245dd126ecfea817108e2cefa4cf02
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8566
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Haiku-Format: Haiku-format Bot <no-reply+haikuformatbot@haiku-os.org>
* Allocate blocks and add them to the hash table so they are
available for a future block_cache_get call.
* Make use of prefetching in FAT driver.
* A client filesystem may request to prefetch a block run that
contains some blocks that are already cached. The request will
be truncated at the first such block in the run.
* Fixes #19186.
Change-Id: I8d2e3cff15e5b46569438e0dc085e2b391aa57a5
Reviewed-on: https://review.haiku-os.org/c/haiku/+/8525
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
