bsdinstall: Improve message for existing FreeBSD EFI entry
When reinstalling FreeBSD bsdinstall reported "There are multiple
FreeBSD EFI boot entries." This sounds like something went wrong in the
past. Clarify that there may be only one existing entry, which is not
surprising for a reinstall.
Reviewed by: manu, ziaee
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D51527
(cherry picked from commit ebc6ff8db17683b566d49fe89a43a668d3d67915)
iwx: enable seqno offload
Enable sequence number offload.
This should both enable the sequence number offloading and disable
the net80211 TX lock from being acquired/released/checked.
Reviewed by: bz, thj
Differential Revision: https://reviews.freebsd.org/D50694
net80211: add support for sequence number offloading
Add support for sequence number offloading -
* Check IEEE80211_CONF_SEQNO_OFFLOAD() before doing TX lock
manipulation;
* Don't call ieee80211_output_seqno_assign() if
IEEE80211_CONF_SEQNO_OFFLOAD() is true.
TODO:
* this doesn't yet do beacon sequence number allocation offloading;
I'll tackle that later.
Differential Revision: https://reviews.freebsd.org/D50692
Reviewed by: bz
amd64 vmx: micro-optimize vmlaunch failure path
Eliminate two unneeded jumps. One is the jmp to the next instruction,
where there is no requrement that vmlaunch is followed by jmp. Another
one conditionally sets %r11d value, and can be replaced by cmovcc.
Reviewed by: markj
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D52136
LinuxKPI: 802.11: factor out sinfo printing
Factor out the sinfo printing into an sbuf from the sysctl handler.
That allows us to use lkpi_80211_dump_lvif_stas() for printing other
debug information as well.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
LinuxKPI: 802.11: improve ieee80211_request_smps()
Adjust logging from TODO() to IMPROVE() now that we can use fmt strings
there too. Make sure we are not getting unsuported values to print.
Also for non-station mode simply return.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
LinuxKPI: 802.11: fill in sinfo for chains in case driver does not
We are getting per-packet information on the rx chains. Save them
locally like we do for rx rate and add them to sinfo for the sysctl
in case the driver does not fill in that information.
Also note: chain_signal is a bitmask and not a number; treat it
accordingly.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
LinuxKPi: 802.11: adjust linuxkpi_set_ieee80211_dev() to set ic_name
While we used to pass in the dev_name(), we now use wiphy_dev() inside
the implementation to get the dev and the name. This is for two reasons:
(1) ath1xk also unset the dev and we need to check once we add support,
and (2) in preparation for future work, which needs to hook into that
point.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
LinuxKPI: 802.11: correct struct linuxkpi_ieee80211_channel
Some of the annotations were not quite right. Also various fields had
not been typed properly. Correct the types of field members and
given the change also re-sort.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
LinuxKPI: 802.11: announce netdev features under boot -v
Under bootverbose print the announced netdev features during
ifattach. Helps us to debug problems amongst drivers if needed.
Leave a comment where to add checksum offlooading setting on the
ifp if we get there (this would be based on netdev features).
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
LinuxKPI: 802.11: improve reg_query_regdb_wmm()
Adjust the IMPROVE macro to also take a format string and use that
in reg_query_regdb_wmm() instead of TODO() to avoid spamming the console
with something that needs larger reworks. Return -ENODATA instead of
-ENXIO as that seems a better fit though the only consumer uses it
without checking the return value these days.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
linuxKPI: 802.11: assert wihpy lock in lkpi_80211_mo_start()
Newer versions of Linux wifi drivers assert the wiphy lock in the
(*start)() path. Make sure we do too and do hold the wiphy lock.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
LinuxKPI: 802.11: add guards to lkpi_ic_ampdu_rx_{start,stop}
After updating calls into lkpi_ic_ampdu_rx_stop() were seen along
with panics.
Working backwards adding checks to each level we lastly ended up fighting
a lsta which is NULL. Simply return in any error cases as there is
nothing we can do and just leave it to net80211.
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
param.h: Bump __FreeBSD_version for 3f61ececacd8
Bump __FreeBSD_version to 1403505 for 3f61ececacd8
since this commit changes the internal KAPI between
the NFS modules.
vfs_syscalls.c: Fix handling of offset args for copy_file_range
Commit 197997a broke handling of the offset
arguments to copy_file_range() when specified non-NULL.
The code fails to update the offsets and, as such, a loop like:
do {
len = copy_file_range(infd, &inpos, outfd, &outpos,
SSIZE_MAX, 0);
} while (len > 0);
becomes an infinite loop, just doing the same copy over and
over again.
This patch fixes it.
The clause "(foffsets_locked || foffsets_set)" in the if is not
actually needed for correctness, but I thought it made the code
[5 lines not shown]
nfscl: Fix GetACL so that it does not want a reply cached
Without this patch, an NFSv4.1/4.2 GetACL RPC requests that
the session cache the reply. In some cases, the reply may
be too large to cache, resulting in a NFS4ERR_X
error from the server.
Since a GetACL is idempotent, disable reply caching for it,
by setting that it can generate a large reply.
Tested against a Linux server with a large ACL on a file.
(cherry picked from commit 51affb7e97161e260f6d99cb0904daf7cc9f51a7)
nfsd: Avoid relocking vnode for NFSv4 Readdir
Without this patch, nfsv4_fillattr() relocks the vnode to
test to see if extended attributes are supported.
This is inefficient and could cause deadlocks if Readdir
ever asks for this attribute.
At this time, no extant NFSv4 client asks for this attribute
for Readdir, but this patch fixes the problem in case a
future client does so, by moving the test for extended attribute
support to before the nfsv4_fillattr() call where the vnode
is still locked.
(cherry picked from commit c00baac0ab46f54951a29b834e064935b749ae9a)
lua: Merge lua 5.4.8
Pick up a few bug fixes from lua 5.4.8:
267ef461 Bug: check for constructor overflow in [exp] fields
d1ee2a4d Bug: Bad error message with fields of string _ENV
3fe7be95 Bug: message handler can be overwritten
983bc433 Bug: Use after free in 'luaV_finishset'
25da574f Bug: 'luaD_seterrorobj' should not raise errors
f5e55be2 Bug: Missing error status in panic function
9f0c0fe0 Bug: Wrong limit for local variables in 16-bit systems
782ef85b Bug: wrong code gen. for indices with comparisons
30982bec Bug: Bad stack manipulation in 'multiline' (REPL)
sys/modules: fix standalone build for multiple modules
Multiple Makefile miss opt_*.h and *_if.h header file. We fix it by
running make in sys/modules to build all modules.
Approved by: lwhsu (mentor), markj (mentor)
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D52062
hkbd: remove erorr detection in KDSKBSTATE ioctl
The KDSKBSTATE ioctl brings the LED up. However, some keyboards (like qemu
keyboard) may not have LED or failed to set the LED due to unexpected reason.
Therefore, removing the error check as ukbd(4) does allow the keyboard works
correctly with kbdcontrol(4).
Also move hw.hid.hkbd.no_leds sysctl out of HID_BUG thus users can
disable setting LEDs
PR: 288968
Reviewed by: wulf
Tested by: trashcan at ellael.org, marklmi26-fbsd at yahoo.com, trkellers at gmail.coom
Approved by: lwsu (mentor), markj (mentor)
MFC after 1 week
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D52101
mvalsmva
libelftc: Move to toolchain package
This is where the rest of ElfToolchain lives now. Note that this
library only installs manpages.
Reported by: des
Differential Revision: https://reviews.freebsd.org/D51899
packages: Improve handling of -lib packages
For some packages (OpenSSL, Kerberos) we want to ship runtime libraries
in a separate package, e.g. openssl and openssl-lib. Currently this is
done using PACKAGE=openssl-lib, but that creates packages with strange
names like openssl-lib-lib32.
Instead, add a new LIB_PACKAGE option to bsd.lib.mk that causes runtime
libraries to be placed in a new -lib subpackage. This significantly
improves the set of packages we create; for example, OpenSSL goes from:
FreeBSD-openssl
FreeBSD-openssl-dbg
FreeBSD-openssl-lib
FreeBSD-openssl-lib-dbg
FreeBSD-openssl-lib-dbg-lib32
FreeBSD-openssl-lib-dev
FreeBSD-openssl-lib-dev-lib32
FreeBSD-openssl-lib-lib32
[19 lines not shown]
Remove redundant PACKAGE for INTERNALLIB libraries
These libraries don't install anything, so they shouldn't have a
PACKAGE setting. This avoids surprising behaviour in future if
e.g. manpages are added to an internal library.
Reported by: des
Differential Revision: https://reviews.freebsd.org/D51901
