IABSD.fr/src/sys/dev
Branch :
Log
| Author | Commit | Date | CI | Message |
|---|---|---|---|---|
 |
2f6a94ad | 2020-07-27 07:24:03 | trim trailing whitespace in if_iwn.c Patch by Holger Mikolon | |
 |
c04334bf | 2020-07-26 17:44:15 | Avoid nvram lock timeout on sparc64 systems with onboard BCM5704 instances that come without a fitted EEPROM/NVRAM. ok deraadt@, kn@ | |
|
a2c1178d | 2020-07-26 12:23:32 | Reduce IWN_DEBUG message noise related to scanning. Patch by Holger Mikolon ok tobhe@ | |
 |
9b4e521a | 2020-07-25 16:34:30 | No need for a bzero(), just init saa.sf to NULL. All the saa.saa fields are explicitly initialized. | |
|
78c5d546 | 2020-07-25 13:50:49 | Nuke unused struct scsi_link member of struct wd33c93_softc. SGI compile tests by visa@ | |
|
500422e9 | 2020-07-24 12:43:31 | Turning on various scsi drivers' *DEBUG options reveals that this has rarely (if ever) been done. Fix many printf format errors to calm clang and gcc on amd64, i386, hppa. Missing #include, complaints if 'option <blah>DEBUG' is used in config files, etc. All in debug code. | |
 |
371cc347 | 2020-07-23 11:09:13 | drm/i915/perf: Use GTT when saving/restoring engine GPR From Umesh Nerlige Ramappa b3944d5248c6be0f2dc95a231822c6c67d6f6fd9 in linux 5.7.y/5.7.10 aee62e02c48bd62b9b07f5e297ecfc9aaa964937 in mainline linux | |
|
fc19d940 | 2020-07-23 11:05:52 | drm/i915/gvt: Fix two CFL MMIO handling caused by regression. From Colin Xu 7924e77bf17003aad76f31126d27afe17439876d in linux 5.7.y/5.7.10 fccd0f7cf4d532674d727c7f204f038456675dee in mainline linux | |
|
12f7ecba | 2020-07-23 11:03:14 | drm/i915/gt: Only swap to a random sibling once upon creation From Chris Wilson b99382de1233866a7eea4a016029d417bdcc0fe7 in linux 5.7.y/5.7.10 110f9efa858f584c6bed177cd48d0c0f526940e1 in mainline linux | |
|
033fc66b | 2020-07-23 11:00:13 | drm/i915/gt: Ignore irq enabling on the virtual engines From Chris Wilson efce4c5bf8873329a578b70c4b5b0df0fee73e2d in linux 5.7.y/5.7.10 858f1299fd6f7518ddef19ddd304c8398ac79fa5 in mainline linux | |
|
d8b0789b | 2020-07-23 10:57:50 | drm/i915: Move cec_notifier to intel_hdmi_connector_unregister, v2. From Maarten Lankhorst bc4727542c394ac99e1826e35d708d83088591d3 in linux 5.7.y/5.7.10 6647e6cdba753e71170be7da2acfead7154f56d8 in mainline linux | |
|
117df390 | 2020-07-23 10:54:50 | drm/amdgpu/display: create fake mst encoders ahead of time (v4) From Alex Deucher 4c76a5a3db25f24accb12acf4a6f3707ee283761 in linux 5.7.y/5.7.10 3168470142e0a82b5732c04ed4c031a9322ae170 in mainline linux | |
|
f0851b17 | 2020-07-23 10:49:23 | drm/amd/display: OLED panel backlight adjust not work with external display connected From hersen wu 702e13f8d26c6d18389c299347580783fef02bcd in linux 5.7.y/5.7.10 b448d30b0c303d5454ea572b772d1ffae96bc6e7 in mainline linux | |
|
332c92d0 | 2020-07-23 10:46:17 | drm/amd/display: handle failed allocation during stream construction From Josip Pavic adfe5ec9ac3ff13e9dabe42216fb34a9e9c19738 in linux 5.7.y/5.7.10 be73e608ae2711dc8a1ab8b9549d9e348061b2ee in mainline linux | |
|
473d62af | 2020-07-23 10:43:10 | drm/amdgpu/sdma5: fix wptr overwritten in ->get_wptr() From Xiaojie Yuan 7c7df36732772d4f68e0ed50667ced18440bb2ad in linux 5.7.y/5.7.10 05051496b2622e4d12e2036b35165969aa502f89 in mainline linux | |
|
2aee0a62 | 2020-07-23 10:40:17 | drm/amdgpu/powerplay: Modify SMC message name for setting power profile mode From chen gong b22a2501543e9ebe88b8375652ad41715f1e765d in linux 5.7.y/5.7.10 98a34cf931e848f8489d3fb15a8f5fc03802ad65 in mainline linux | |
 |
f94c830f | 2020-07-23 10:10:15 | Even more whitespace fixup. | |
|
6d00698e | 2020-07-23 09:21:51 | match on AMDI0010 Makes imt(4) and ims(4) attach and touchpad work on jmc@'s Inspiron 5505. "pms0: Elantech Clickpad, version 4, firmware 0x5f1001" also probes but by itself did not result in a functional touchpad. ok kettenis@ | |
 |
1d772be3 | 2020-07-23 09:17:03 | Fix comments to match code; OK fcambus@ | |
 |
e3eac33c | 2020-07-23 01:24:53 | Increase the event queue size. When polling for admin command completion, we don't wait for the event to be posted to the queue, we just look at the command itself, which means we can build up a backlog of events to be posted. Newer firmware for ConnectX-4 seems to get upset if the backlog grows beyond some fraction of the event queue size, causing an interrupt storm. This was reported by patrick@ and Hrvoje Popovski (at least) while testing support for multiple tx/rx queues. With the new event queue size, we can safely create 8 queues. | |
|
09e5d825 | 2020-07-22 21:14:05 | Init GOP and COMPHY. | |
|
fad015d2 | 2020-07-22 20:58:40 | Pass tid we're looking for to mvpp2_prs_hw_read(), then we can zero the prs entry there and also save a few lines. | |
|
fef01367 | 2020-07-22 20:50:16 | Rework mvpp2_prs_flow_find() to return a tid instead of a whole prs entry, saving at least one unnecessary malloc(9)/free(9) cycle. | |
|
47b8e6a0 | 2020-07-22 19:57:59 | Fix more whitespace issues. | |
|
e042150e | 2020-07-22 19:56:42 | Use correct ethertype and IP proto defines. | |
|
70ad0a12 | 2020-07-22 19:54:05 | Fix whitespace issue. | |
|
af1244a6 | 2020-07-22 19:53:11 | Use 1U for BIT(x), always define it and use it in four more places. | |
|
7380a3a4 | 2020-07-22 13:16:04 | Nuke unused struct scsi_link members of adapter softc's where the driver successfully compiles on one or more of amd64, i386, hppa. | |
|
3941079e | 2020-07-22 11:13:20 | regen | |
|
893f5127 | 2020-07-22 11:12:40 | add Intel AX200 Bluetooth | |
|
5394f138 | 2020-07-22 10:45:03 | regen | |
|
4910c09b | 2020-07-22 10:43:14 | add devices found in Inspiron 5505 dmesg from jmc@ | |
 |
23293512 | 2020-07-22 02:16:01 | deprecate interface input handler lists, just use one input function. the interface input handler lists were originally set up to help us during the intial mpsafe network stack work. at the time not all the virtual ethernet interfaces (vlan, svlan, bridge, trunk, etc) were mpsafe, so we wanted a way to avoid them by default, and only take the kernel lock hit when they were specifically enabled on the interface. since then, they have been fixed up to be mpsafe. i could leave the list in place, but it has some semantic problems. because virtual interfaces filter packets based on the order they were attached to the parent interface, you can get packets taken away in surprising ways, especially when you reboot and netstart does something different to what you did by hand. by hardcoding the order that things like vlan and bridge get to look at packets, we can document the behaviour and get consistency. it also means we can get rid of a use of SRPs which were difficult to replace with SMRs. the interface input handler list is an SRPL, which we would like to deprecate. it turns out that you can sleep during stack processing, which you're not supposed to do with SRPs or SMRs, but SRPs are a lot more forgiving and it worked. lastly, it turns out that this code is faster than the input list handling, so lots of winning all around. special thanks to hrvoje popovski and aaron bieber for testing. this has been in snaps as part of a larger diff for over a week. | |
 |
81a23f33 | 2020-07-22 00:48:02 | Add support for newer RTL8125 chipset (RTL8125B). Tested for on a TP-LINK TL-NG421. ok jmatthew@ | |
 |
ad7f7af0 | 2020-07-21 03:48:04 | acpi can use IPL_BIO (a low interrupt) since it only enqueues operations for later processing. The use of a high interrupt will predate suspend/resume efforts, we had to redesign acpi to be non-reentrant obviously discussed with kettenis, in snaps for more than a week | |
|
20490285 | 2020-07-20 20:38:10 | Neither is saa.ssa.pool the same as saa.saa_pool. | |
|
7fc4ef32 | 2020-07-20 16:09:48 | Missed a chunk somewhere along the line. Use our_id vs our_id_b in appropriate spots. Don't copy one empty/unused scsi_link (channel) to another empty/unused scsi_link (channel_b) | |
|
e5eae15d | 2020-07-20 14:41:12 | Move remaining scsi bus initialization info from "prototype scsi link" fields to struct scsibus_attach_args. Nuke the struct scsi_link * (saa_sc_link) in scaibus_attach_args. Explicitly initialize each field in scsibus_attach_args variables. | |
 |
e5e6042b | 2020-07-20 13:34:31 | Update Spleen kernel fonts to version 1.8.1, bringing the following improvements: - Center tilde vertically for all sizes, as modern fonts do | |
|
63cfed36 | 2020-07-20 12:40:45 | Fix various whitespace and formatting issues in rasops(9). | |
|
23f86de9 | 2020-07-20 08:22:06 | KNF for multi-line comment added to iwn5000_set_gains() in previous commit. | |
|
5ed5d284 | 2020-07-20 08:19:59 | Fix gain calibration for some iwn(4) devices (5000 and up). IWN_LSB() returns an index starting with 1, however the arrays used later on (noise and gain in iwn5000_set_gains()) start with 0. The previous code accounted for this difference when setting the antenna gain by accessing cmd.gain[i - 1]. However the noise array was accessed with noise[i], the chainmask was checked against i, and more importantly the overall for() loop iterated wrongly over the antennas by always starting with i=2 (the third antenna). One consequence is that gain calibration never happened in case of only two antennas. Patch by Holger Mikolon. | |
|
5c60d790 | 2020-07-20 08:09:30 | Fix a logic bug in iwn(4) which meant that automatic rate control for A-MPDU ran while a fixed Tx MCS was configured. The intention was of course the inverse: Use automatic rate control if Tx MCS is not fixed. There is a second bug which went unnoticed because of the above bug. The value of 'seq' provided by firmware may be smaller than ba_winstart. When this happened it triggered a KASSERT in ieee80211_output_ba_record_ack. iwn_ampdu_rate_control() implicitly assumes that its 'seq' parameter equals ba_winstart, so just pass that value directly. tested by bket@, jmc@, and myself ok mpi@ | |
|
d32efed9 | 2020-07-20 08:04:41 | update a comment in iwn_ampdu_tx_done following previous SSN -> SEQ change | |
|
ffe363d1 | 2020-07-20 08:00:38 | Make iwn_ampdu_tx_done() record an ACK for the frame for which the hardware triggered the Tx completion interrupt, instead of the frame with the starting sequence number (SSN) of the firmware's block ack window. The frame at SSN is unrelated and may not even have been transmitted yet. tested by bket@, jmc@, and myself ok mpi@ | |
|
eb9f82c0 | 2020-07-20 07:45:44 | The IEEE80211_F_HIDENWID flag is now part of ic_userflags, not ic_flags. Fix code which was still looking for this flag at the old location. The 'hidenwid' feature was slightly broken as a result: The SSID was leaked in probe responses to wildcard probe requests. There are other trivial ways of snooping a "hidden" SSID however so this is not a big deal. Problem reported by Mogens Jensen. | |
|
14655f60 | 2020-07-20 03:15:56 | use timeout_del_barrier(9) for del_timer_sync() | |
|
ead808c4 | 2020-07-19 18:57:57 | Move the adapter related items (luns, adapter, adapter_target, adapter_buswidth, adapter_softc) from struct scsi_link to struct scsibus_attach_args. Additional compile tests by jmatthew@ (sparc64) and aoyam@ (luna88k). | |
 |
03d30443 | 2020-07-19 00:26:40 | regen | |
|
b453b57a | 2020-07-19 00:26:00 | Add a few more Atheros ids; ok deraadt | |
|
060e6ef1 | 2020-07-18 20:21:01 | Make /dev/ksyms work. | |
|
1a409cc7 | 2020-07-18 09:44:59 | Add functions to set/add properties on an FDT node. ok visa@ | |
|
2cacaaf4 | 2020-07-18 07:18:22 | replace stat collection with the kstat api. before this the stats were always collected, but you could only see stats with gdb against the running kernel, or by enabling IX_DEBUG in the build. neither of these are supported by default. | |
|
452daaed | 2020-07-17 08:07:33 | Re-work intr_barrier(9) on arm64 to remove layer violation. So far we have stored the struct cpu_info * in the wrapper around the interrupt handler cookie, but since we can have a few layers inbetween, this does not seem very nice. Instead have each and every interrupt controller provide a barrier function. This means that intr_barrier(9) will in the end be executed by the interrupt controller that actually wired the pin to a core. And that's the only place where the information is stored. ok kettenis@ | |
|
7177b505 | 2020-07-17 07:58:46 | Merge the two if blocks checking if the font in use contains a given character in rasops_mapchar(). OK kn@, cheloha@ | |
 |
b8362a7b | 2020-07-17 07:53:58 | Revert r1.170. dlg and jmatthew simultaneously fixed this the correct way. | |
|
2405a0d3 | 2020-07-17 07:49:49 | i forgot to commit this with if_ix.c r1.169 stoeplitz code needs to #include <net/toeplitz.h>, and ix does it's includes here. pointed out by jmatthew@ (thanks) | |
|
52648260 | 2020-07-17 07:40:35 | This needs <net/toeplitz.h> for stoeplitz_to_key(9). | |
|
a522f7e1 | 2020-07-17 06:33:07 | use the system stoeplitz key instead of a random one. while here, stash the rss hash value from rx descriptors in the mbuf as a flow id. | |
|
a4cd2d76 | 2020-07-17 06:27:36 | enable multiq support for ix. this is an updated version of a diff from christiano haesbaert by way of mpi@ to enable the use of multiple tx and rx rings with msi-x. now that patrick@ got interrupts on multiple cpus working on arm64, i can commit this. the main changes are to use an intrmap pointer as the flag for whether msix has been enabled or not, and to use the intrmap to establish vectors on multiple cpus. tested by hrvoje popovski | |
|
3ac368de | 2020-07-17 03:37:37 | Virtual functions are effectively identical to full physical functions, so we can attach to them too. ok dlg@ | |
|
664b6d32 | 2020-07-17 03:37:36 | name the rx rings so systat mb shows them. | |
|
a59da6ce | 2020-07-17 03:23:18 | Consistently use the port type and speed register (PTYS) to determine if the link is up, rather than the operational status (PAOS). ok dlg@ | |
|
2caca9d6 | 2020-07-17 02:58:14 | add kstats to myx. myx is unusually minimal, so there's not a lot of information that the chip provides. the most interesting is the number of packets the chip drops cos of a lack of space on the rx rings. | |
|
499fa1f2 | 2020-07-17 00:36:42 | Always a joker hiding somewhere. Access adapter_buswidth via link->bus->sb_adapter_buswidth. | |
|
604cf95e | 2020-07-16 21:49:41 | Pass the interrupt handler cookie instead of the pointer to it to intr_barrier(9). Fixes mysterious panics seen while working on intr_barrier(9) for arm64. ok jmatthew@ | |
|
0b29cb40 | 2020-07-16 21:18:29 | Access adapter softc via link->bus->sb_adapter_softc. In sparc64 autoconf access 'luns' via sb_luns and 'adapter_buswidth' via sb_adapter_buswidth. Removes last post-config uses of the copies of bus related information in scsi_link. | |
|
2d8fcdc3 | 2020-07-16 10:17:09 | drm/i915: Also drop vm.ref along error paths for vma construction From Chris Wilson 5e53344673fcf1df8ff52675ab0539fff26a8e29 in linux 5.7.y/5.7.9 cf1976b11372cac3b57fbae1831f66a4486355d3 in mainline linux | |
|
ee7f71c2 | 2020-07-16 10:14:13 | drm/i915: Drop vm.ref for duplicate vma on construction From Chris Wilson 98762e5b00567622d57295d2d3389aea2c37be03 in linux 5.7.y/5.7.9 42723673a193d5f8e30dba6ea9826d42262a502b in mainline linux | |
|
28304a26 | 2020-07-16 10:11:33 | drm/amdgpu: asd function needs to be unloaded in suspend phase From Huang Rui 22ff658396b44606163ad0231e8215309a26d613 in linux 5.7.y/5.7.9 20303ec5d2165ee6344190274bc59118921f71d9 in mainline linux | |
|
6bd9a77d | 2020-07-16 10:09:22 | drm/amdgpu: add TMR destory function for psp From Huang Rui 2c41c968c6f6480860c67210815cadc6507f5014 in linux 5.7.y/5.7.9 c564b8601ae917086751d90f464d5f19d731ece7 in mainline linux | |
|
60ed5e11 | 2020-07-16 10:07:02 | drm/amdgpu: don't do soft recovery if gpu_recovery=0 From Marek Olsak d7915047a1d59474d06789a9dd1925064969e080 in linux 5.7.y/5.7.9 f4892c327a8e5df7ce16cab40897daf90baf6bec in mainline linux | |
|
ec9a754d | 2020-07-16 10:03:44 | drm/i915: Skip stale object handle for debugfs per-file-stats From Chris Wilson d073ed965cd908d320d49824dee02c80168f96a8 in linux 5.7.y/5.7.9 7dfbf8a07cf8c936b0d6cc810df6ae7923954d5b in mainline linux | |
|
15914761 | 2020-07-16 10:01:41 | drm/i915/gt: Pin the rings before marking active From Chris Wilson 096b85c64987a0ef4fcc6ac0e3c3661573828df1 in linux 5.7.y/5.7.9 5a383d443b29a140094430f3ad1d02fa1acc2b80 in mainline linux | |
|
82efeb8e | 2020-07-16 09:59:17 | drm/radeon: fix double free From Tom Rix 67e9648f20524ff6824ebeb26afea45f7d9c300d in linux 5.7.y/5.7.9 41855a898650803e24b284173354cc3e44d07725 in mainline linux | |
|
df64a991 | 2020-07-16 09:56:50 | drm: panel-orientation-quirks: Use generic orientation-data for Acer S1003 From Hans de Goede 109b3b4e2a0f5bfda023a1b91b404fac31ba6808 in linux 5.7.y/5.7.9 a05caf9e62a85d12da27e814ac13195f4683f21c in mainline linux | |
|
86b21ffc | 2020-07-16 09:53:49 | drm: panel-orientation-quirks: Add quirk for Asus T101HA panel From Hans de Goede e64e92fea2bf85d9f33304faf973c46859afd0f7 in linux 5.7.y/5.7.9 6c22bc18a3b93a38018844636557ad02e588e055 in mainline linux | |
|
ff2b09e8 | 2020-07-16 09:50:39 | drm/ttm: Fix dma_fence refcnt leak when adding move fence From Xiyu Yang bfd4297fd74ebb18d05044e49f1ff2fedd2ece8d in linux 5.7.y/5.7.9 11425c4519e2c974a100fc984867046d905b9380 in mainline linux | |
|
9eae3c38 | 2020-07-16 09:48:22 | drm/ttm: Fix dma_fence refcnt leak in ttm_bo_vm_fault_reserved From Xiyu Yang f5e6ec57e3cbe37cd3ef52d18de6be9953ee105a in linux 5.7.y/5.7.9 37cc4b95d13f311c04aa8e9daacca3905ad45ca7 in mainline linux | |
|
aff45b01 | 2020-07-16 05:45:03 | Remove the refill timeout and wait for interrupts to finish when bringing the interface down. prompted by a crash seen by Hrvoje Popovski ok dlg@ | |
|
8014a50b | 2020-07-16 03:04:50 | use a mutex to serialise the test and set of ifp->if_link_state. this was serialised by NET_LOCK, but now i get link state change information in an interrupt context, so i shouldn't (can't) do that anymore. ok jmatthew@ | |
|
4e804522 | 2020-07-16 00:58:02 | sc_atq_mtx is unused, so get rid of it | |
|
b296e835 | 2020-07-16 00:52:33 | Initialize sis_ring_init() 'sis_rx_prod' and 'sis_rx_cons' to 0. ok mpi@, deraadt@ | |
 |
24c7ba2a | 2020-07-15 11:33:12 | Run the sxitemp(4) at IPL_SOFTCLOCK instead of IPL_VM. Prevents "panic: mtx ...: locking against myself" on Orange Pi Zero. Analysis by patrick@: "The thermal sensor framework uses its own taskq with IPL_SOFTCLOCK. sxitemp(4) calls thermal_sensor_update() from interrupt context, and sxitemp(4) is using IPL_VM (memory allocation?!) for its interrupt. IPL_VM is obviously higher than IPL_SOFTCLOCK, so it ends up being able to interrupt the taskq. Even though we're in msleep_nsec, I think we have *not yet* given up the mutex, that we are holding while looking for more work, only releasing it while sleeping. Thus, the interrupt runs task_add(), which tries to grab the taskq's mutex, even though the taskq already holds it!" ok patrick@ kettenis@ | |
|
34a5437d | 2020-07-15 08:34:25 | Optimize character rendering in 32bpp mode, with unrolled rows and pairwise foreground / background pixel rendering. This does double-pixel rendering for the common font widths (8, 12, 16, and 32 pixels wide), resulting in a significant speed-up. No regression noticed on 32-bit architectures. Tested by weerd@, and by abieber@ on macppc. From John Carmack, thanks! | |
|
ca19ff74 | 2020-07-15 04:36:02 | patrick@ pointed out there was a gap in the flow group numbering. ok dlg@ | |
|
11623c6a | 2020-07-15 03:32:51 | Consistently convert syndrome codes from big endian before printing them. ok dlg@ | |
|
d67371fd | 2020-07-14 15:42:19 | Implement pci_intr_establish_cpu() on arm64 and armv7. The function pointer in the chipset tag for establishing interrupts now takes a struct cpu_info *. The normal pci_intr_establish() macro passes NULL as ci, which indicates that the primary CPU is to be used. The PCI controller drivers can then simply pass the ci on to our arm64/armv7 interrupt establish "framework". Prompted by dlg@ ok kettenis@ | |
|
789e88a4 | 2020-07-14 15:34:14 | Extend the interrupt API on arm64 and armv7 to be able to pass around a struct cpu_info *. From a driver point of view the fdt_intr_establish_* API now also exist same functions with a *_cpu suffix. Internally the "old" functions now call their *_cpu counterparts, passing NULL as ci. NULL will be interpreted as primary CPU in the interrupt controller code. The internal framework for interrupt controllers has been changed so that the establish methods provided by an interrupt controller function always takes a struct cpu_info *. Some drivers, like imxgpio(4) and rkgpio(4), only have a single interrupt line for multiple pins. On those we simply disallow trying to establish an interrupt on a non-primary CPU, returning NULL. Since we do not have MP yet on armv7, all armv7 interrupt controllers do return NULL if an attempt is made to establish an interrupt on a different CPU. That said, so far there's no way this can happen. If we ever gain MP support, this is a reminder that the interrupt controller drivers have to be adjusted. Prompted by dlg@ ok kettenis@ | |
|
24a3fa5d | 2020-07-14 11:22:11 | reduce the diff to linux in local versions of functions brings in changes from 'drm/ttm: stop always moving BOs on the LRU on page fault' 5d50fcbda7b0acd301bb1fc3d828df0aa29237b8 and some other minor changes ok kettenis@ | |
|
07af7ebb | 2020-07-14 04:10:18 | Set up a second level flow table for RSS and switch the root flow table entries to forwarding packets to it. The RSS flow table contains flow table entries that match specific types of traffic and send them to TIRs set up to hash on the appropriate fields, then deliver packets through an RQT to all our rx queues. We still only have one queue, but now all we need to do is plug in an intrmap to add more. ok dlg@ | |
|
f226851b | 2020-07-14 00:43:29 | Set SDEV_NOWIDE and SDEV_NOSYNC on the 'real' link (xs->sc_link) when negotiations fail. | |
|
f1e1e8b4 | 2020-07-13 16:23:53 | Enable background scanning on iwx(4) devices. | |
|
50b95bf0 | 2020-07-13 16:18:52 | Increase iwx(4) command queue size to the size expected by hardware. The iwx(4) command queue has 256 entries in hardware. The Linux driver contains an abstraction layer which allows it to allocate just 32 command queue entries in software. Because I misunderstood how this works iwx(4) ended up with 32 command queue entries, but without a shim which presents these 32 entries as 256 entries to hardware. Our command queue size was later bumped to 64 to make fatal firmware errors during initialization disappear. The problem was still not fully understood at that time. I didn't realize that firmware errors coincided with the command queue index wrapping around. Hardware expects the queue to wrap at 256 entries so fix this problem for good in a simple way: Allocate 256 entries for the command queue, just like iwm(4) does. This will allow us to enable background scanning on iwx(4), which so far triggered firmware errors due to additional scan commands causing the command queue index to wrap at 64. | |
|
d114bbad | 2020-07-13 14:25:41 | Where we have a pointer (sc) to the relevant adapter_softc, use &sc->sc_dev to get a device pointer rather than using the void * whence we obtained 'sc'. | |
|
d4769f94 | 2020-07-13 13:43:31 | Remove some pointless casting of void * when assigning one (scsi_link.adapter_softc) to a pointer of a particular softc type. | |
|
48439b74 | 2020-07-13 10:40:03 | More preparations for adding RSS. Adjust several functions to allow for the creation of more flow tables and TIRs, and more types of flow entries and TIRs, consolidate flow group details into a single struct, and improve the way doorbells in host memory are laid out. The doorbell memory area is divided into halves for CQs and SQ/RQs, with the doorbells for each CQ and SQ/RQ pair kept 64 bytes apart. This arrangement will allow for up to 32 queues, which is plenty for now. ok dlg@ | |
|
9a82318d | 2020-07-13 10:35:55 | add kstat support for reading hardware counters. this replaces the existing counters implementation, which just collected the stats in the softc, but didn't really provide a way for a person to read them. em counters get cleared on read. a lot of them are 32bit, so to avoid overflow the counters are polled and the newly accumulated values are added to some 64 bit counters in software. tested by hrvoje popovski and SAITOH Masanobu ok mpi@ i missed these files when i committed src/sys/dev/pci/if_em.c r1.356. thanks to jsg for pointing this out. | |
|
86e3bb8a | 2020-07-13 08:31:32 | When athn(4) is deciding how to decrypt an incoming frame, look into the ieee80211_node data structure to find the WPA group cipher which was negotiated with the peer, rather than checking the wireless interface's configuration data. Found while investigating a problem where WPA2+TKIP wasn't working with athn. Problem reported by Tim Chase. | |
|
63b35fb2 | 2020-07-13 06:25:17 | change some drm locks from IPL_TTY to IPL_NONE In drm linux spinlocks are mapped to mutex(9). Locks without calls to spin_lock_irqsave(), spin_lock_irq() and the like (which block interrupts) can be changed to IPL_NONE. ok kettenis@ |