ĵ225503 171 158 257 283 521 252 228 503 234 102 231 236 225 348 727 203 419 633 550 1140 293 203 165 209 212 317 266 487 941 263 209 259 302 274 181 662 301 361 126 133 391 370 232 305 151 144 1540 286 109 200 184 110 191 123 433 199 192 306 232 233 234 130 160 486 252 211 272 142 455 226 310 488 204 365 217 174 214 214 603 165 330 393 233 245 383 267 121 112 188 165 230 157 930 930 275 223 174 847 390 1028 256 167 1749 516 283 374 182 363 343 113 242 218 210 131 158 653 178 245 241 201 208 278 263 588 211 271 255 305 282 150 161 228 131 248 184 343 105 99 382 461 235 329 190 1447 392 776 242 134 173 134 467 333 349 133 253 592 1220 189 153 168 360 159 317 130 105 133 276 242 188 393 255 319 K 10 svn:author V 5 dougb K 8 svn:date V 27 2011-09-12T04:56:48.059963Z K 7 svn:log V 64 s/printf/echo -e/ for consistency, and use it to full advantage END K 10 svn:author V 7 delphij K 8 svn:date V 27 2011-09-12T06:41:13.600165Z K 7 svn:log V 160 MFC r224002: Add a new option, -N to disable the default and recommended syslogd(8) behavior, which binds to the well known UDP port. This option implies -s. END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-12T08:38:21.851684Z K 7 svn:log V 190 dsp_ioctl: fix type of variable used to store ioctl request PR: kern/156433 Submitted by: Grigori Goronzy Reviewed by: hselasky Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 2 ed K 8 svn:date V 27 2011-09-12T10:07:21.330147Z K 7 svn:log V 429 Fix error return codes for ioctls on init/lock state devices. In revision 223722 we introduced support for driver ioctls on init/lock state devices. Unfortunately the call to ttydevsw_cioctl() clobbers the value of the error variable, meaning that in many cases ioctl() will now return ENOTTY, even though the ioctl() was processed properly. Reported by: Boris Samorodov Patch by: jilles@ Approved by: re@ (kib@) END K 10 svn:author V 7 attilio K 8 svn:date V 27 2011-09-12T11:18:54.926039Z K 7 svn:log V 155 MFC r225372: Disable the watchdog while a KDB backend is running and kick it again just before to return KDB context. Sponsored by: Sandvine Incorporated END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-12T14:21:06.957858Z K 7 svn:log V 132 It's wasteful to only check the active txqs from that bitmask; we may as well check all of them at onc. Obtained from: Linux ath9k END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-12T14:23:27.915331Z K 7 svn:log V 407 HAL_INT_GLOBAL needs to keep being set here so SWBA can fire appropriately whilst the rest of the interrupts are disabled. This constant rewriting of the interrupt mask/enable registers seems a bit silly to me; I'll have to tidy it up at some point. Linux just toggles AR_IER (global interrupt enable) on/off - but it does deferred beacon processing, rather than doing it in interrupt context like we do. END K 10 svn:author V 7 pluknet K 8 svn:date V 27 2011-09-12T14:32:35.985839Z K 7 svn:log V 137 Add missing lib32 parts for the previous changes: removal of Objective-C support and libpkg; libsbuf version bump Approved by: re (kib) END K 10 svn:author V 7 attilio K 8 svn:date V 27 2011-09-12T14:39:03.559710Z K 7 svn:log V 7 MFCMFC END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-12T15:01:32.642016Z K 7 svn:log V 135 Fix some debugging - don't deference tap if it's NULL. This will occur on non-11n NICs, like the AR5212 I just tried this codebase on. END K 10 svn:author V 3 kib K 8 svn:date V 27 2011-09-12T15:11:34.845708Z K 7 svn:log V 143 Use atomics to update the page dirty field, in a way similar to the handling of the aflags. Disscussed with: alc, attilio (previous versions) END K 10 svn:author V 3 kib K 8 svn:date V 27 2011-09-12T15:13:23.913513Z K 7 svn:log V 132 After the dirty updates are atomic, there is no need to lock page queues only to guarantee proper dirty updates. Submitted by: alc END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-12T15:21:52.393490Z K 7 svn:log V 255 Partially revert 222753: If a CardBus card stores its CIS in a BAR, delete the BAR after parsing the CIS. This forces the resource range to be reallocated if the BAR is reused by the device. Submitted by: deischen Reviewed by: imp Approved by: re (kib) END K 10 svn:author V 7 attilio K 8 svn:date V 27 2011-09-12T20:39:31.401060Z K 7 svn:log V 630 dump_write() returns ENXIO if the dump is trying to be written outside of the device boundry. While this is generally ok, the problem is that all the consumers handle similar cases (and expect to catch) ENOSPC for this (for a reference look at minidumpsys() and dumpsys() constructions). That ends up in consumers not recognizing the issue and amd64 failing to retry if the number of pages grows up during minidump. Fix this by returning ENOSPC in dump_write() and while here add some more diagnostic on involved values. Sponsored by: Sandvine Incorporated In collabouration with: emaste Approved by: re (kib) MFC after: 10 days END K 10 svn:author V 8 brueffer K 8 svn:date V 27 2011-09-12T20:57:22.420672Z K 7 svn:log V 105 Connect the vxge(4) module to the i386/amd64 build. Catcher of stupid errors: kib Approved by: re (kib) END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-12T21:09:56.320035Z K 7 svn:log V 326 Allow the ipfw.ko module built with a kernel to honor any IPFIREWALL_* options defined in the kernel config. This more closely matches the behavior of other modules which inherit configuration settings from the kernel configuration during a kernel + modules build. Reviewed by: luigi Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-12T23:52:55.648815Z K 7 svn:log V 540 - Fix a bug that can lead to displaying an incorrect value. (r224210) - Fix an abnormal termination caused by twice of "rtadvctl disable". (r224303) - Use poll() to wait for the control message socket instead of a spin loop. (r224304) - s/cmsg_/cm_/ to avoid conflict with CMSG_* symbols for struct cmsghdr. (r224619) - Ignore an interface that never sent RAs for graceful shut-down. (r224620) - Refine log messages. (r225148) - Fix SIGSEGV when receiving RAs that contain RDNSS and/or DNSSL options. (r225149) Approved by: re (kib) END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-12T23:55:23.020666Z K 7 svn:log V 457 - Add ":origin" label to the interface id for resolvconf(8). (r223149) - Add -u option to enable adding :[RA-source-address] to the interface id. (r223149) - s/INET6_ADDRSTRLEN/sizeof(ntopbuf)/ (r223149) - Fix a bug that can prevent -D from being overridden by -d. (r223149) - /-P pidfile/-p pidfile/ for consistency with rtadvd(8). (r223149) - Fix -F flag handling to support receiving RAs even when ip6.forwarding=1. (r225150) Approved by: re (kib) END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-13T00:06:11.612941Z K 7 svn:log V 1046 Add $ipv6_cpe_wanif to enable functionality required for IPv6 CPE (r225485). When setting an interface name to it, the following configurations will be enabled: 1. "no_radr" is set to all IPv6 interfaces automatically. 2. "-no_radr accept_rtadv" will be set only for $ipv6_cpe_wanif. This is done just before evaluating $ifconfig_IF_ipv6 in the rc.d scripts (this means you can manually supersede this configuration if necessary). 3. The node will add RA-sending routers to the default router list even if net.inet6.ip6.forwarding=1. This mode is added to conform to RFC 6204 (a router which connects the end-user network to a service provider network). To enable packet forwarding, you still need to set ipv6_gateway_enable=YES. Note that accepting router entries into the default router list when packet forwarding capability and a routing daemon are enabled can result in messing up the routing table. To minimize such unexpected behaviors, "no_radr" is set on all interfaces but $ipv6_cpe_wanif. Approved by: re (bz) END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-13T00:09:47.521286Z K 7 svn:log V 200 - Add an warning when ifconfig_IF_ipv6 has no inet6 keyword in front of an IPv6 address. (r225489) - Use eval for ${ifconfig_args} to fix an issue fixed in r223506. (r225489) Approved by: re (bz) END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-13T02:41:09.449465Z K 7 svn:log V 110 Import of openresolv 3.4.4. Obtained from: http://roy.marples.name/projects/openresolv Approved by: re (kib) END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-13T02:46:22.883941Z K 7 svn:log V 73 Merge from openresolv 3.4.4 in the vendor branch. Approved by: re (kib) END K 10 svn:author V 8 kientzle K 8 svn:date V 27 2011-09-13T05:52:34.522635Z K 7 svn:log V 111 Fix cpio on ARM. PR: bin/160430 Submitted by: Ian Lepore Approved by: re (Kostik Belousov) MFC after: 7 days END K 10 svn:author V 3 kib K 8 svn:date V 27 2011-09-13T08:49:52.645837Z K 7 svn:log V 119 MFC r225416: Add -t switch for mdmfs to enable TRIM on the configured filesystem. While there, fix minor style issues. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-13T09:32:48.948259Z K 7 svn:log V 221 Remove this bit of debugging; it's no longer needed as I think I've finally found the source of the missing TX interrupts. (I'll fix that in a subsequent commit once I've verified the test actually does finally fix it.) END K 10 svn:author V 7 pluknet K 8 svn:date V 27 2011-09-13T13:56:18.017314Z K 7 svn:log V 169 Update ObsoleteFiles.inc: - liblwres shared library version bumped with bind98 import - named.reload removed in favor of rc.d Approved by: re (kib) No objection: dougb END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-13T14:01:35.116444Z K 7 svn:log V 394 zfstest: cleanup the code, improve functionality and diagnostics The utility is not connected to the build, so it should be safe to update it. To do: move the utility to tools/. Some code is provided by Peter Jeremy Tested by: Sebastian Chmielewski , Peter Jeremy (earlier versions) Approved by: re (kib) MFC after: 4 days END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-13T14:03:55.885294Z K 7 svn:log V 848 gpt/zfs boot blocks: reduce optimizing CFLAGS to -O1 gpt and zfs boot blocks are not nearly as size-constrained as boot2 from which they inherited their current optimization and anti-optimization options. As such the current options do not provide any benefit, but make debugging of the code much harder. Also, it has been demonstrated that combination of -mrtd and -fno-unit-at-a-time may result in mis-compilation of the boot code with the current base gcc. Additionally, intermediate assembly file filtering is removed for zfsboot. The new boot blocks are all compile- and boot- tested using qemu. gptzfsboot is tested with real hardware. Reported by: Peter Jeremy [miscompilation] Discussed with: bde, jhb Tested by: Sebastian Chmielewski [gptzfsboot] Approved by: re (kib) MFC after: 3 weeks END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-13T14:07:05.510917Z K 7 svn:log V 170 zfs boot subroutines: correctly specify type of an integer literal Found by adding more warning flags to zfs boot blocks build. Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 7 davidch K 8 svn:date V 27 2011-09-13T15:49:28.018859Z K 7 svn:log V 112 - Fix compiler warning in ADD_64() macro. Approved by: re Obtained from: dimitry@andic.com MFC after: One week END K 10 svn:author V 8 brueffer K 8 svn:date V 27 2011-09-13T15:57:29.175216Z K 7 svn:log V 161 Improve the sleep_delay sysctl description by specifying which unit the number is in. PR: 159975 Submitted by: gcooper Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 3 kib K 8 svn:date V 27 2011-09-13T20:16:11.637602Z K 7 svn:log V 209 Do not try to change the mode or ownership of the root of the mountpoint when newly established mdmfs mount is readonly. PR: bin/128427 Tested and reviewed by: jchandra MFC after: 1 week Approved by: re (bz) END K 10 svn:author V 8 brueffer K 8 svn:date V 27 2011-09-13T20:33:27.463587Z K 7 svn:log V 176 Fix csup to allow case insensitive server names in the auth file, just as advertised in the manpage. PR: 158652 Submitted by: stephen Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 8 brueffer K 8 svn:date V 27 2011-09-13T20:35:34.556827Z K 7 svn:log V 84 Fix typos in error messages. Found by: kib Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 8 rmacklem K 8 svn:date V 27 2011-09-13T21:01:26.454258Z K 7 svn:log V 564 Modify vfs_register() to use a hash calculation on vfc_name to set vfc_typenum, so that vfc_typenum doesn't change when file systems are loaded in different orders. This keeps NFS file handles from changing, for file systems that use vfc_typenum in their fsid. This change is controlled via a loader.conf variable called vfs.typenumhash, since vfc_typenum will change once when this is enabled. It defaults to 1 for 9.0, but will default to 0 when MFC'd to stable/8. Tested by: hrs Reviewed by: jhb, pjd (earlier version) Approved by: re (kib) MFC after: 1 month END K 10 svn:author V 5 gavin K 8 svn:date V 27 2011-09-13T21:15:23.353623Z K 7 svn:log V 206 Include descriptions for keymaps missing from INDEX.keymaps. This fixes the five keymaps referred to only by filename on the keymap selection screen of bsdinstall. Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 8 brueffer K 8 svn:date V 27 2011-09-13T21:39:26.606821Z K 7 svn:log V 263 Allow WPA PSKs up to 63 characters, the max length according to wpa_supplicant.conf(5). Up to now, the PSK was truncated to 15 characters (the length of the input field). PR: 159559 Submitted by: Gavin Chappell Approved by: re (kib) END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-13T23:23:08.496978Z K 7 svn:log V 32 - Fix typo: s/xzefgrep/xzegrep/ END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-13T23:28:04.280618Z K 7 svn:log V 39 - WCS dot position != MBS dot position END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-14T00:40:00.550978Z K 7 svn:log V 296 - Failing fixncomp() does not always mean the pattern is invalid, e.g. MBS with REG_ICASE cannot be handled by the fast matcher. Furthermore, fastncomp() always catches fix patterns so the distinction is not really necessary, it just makes the code longer. Reported by: aakuusta@gmail.com END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-14T01:45:12.558351Z K 7 svn:log V 277 - Document $ipv6_cpe_wanif. - Emphasize $ipv6_enable and $ipv6_prefer are deprecated. - Add more detail descriptions about $ipv6_activate_all_interfaces. - Add some more examples of $ifconfig_IF_ipv6. - rtsold(8) and rtadvd(8) can be used even when ipv6_gateway_enable=NO now. END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-14T01:46:58.597548Z K 7 svn:log V 139 - Document inet6 no_radr flag. - Add descriptions of sysctl(8) variables which can control the default configuration of the inet6 flags. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-14T03:18:00.787675Z K 7 svn:log V 209 Back out the interrupt handling changes for now. There are issues with this in hostap mode leading to lost TX completion events. I'll likely look at merging in a fixed version of this patch at a later stage. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-14T03:25:02.547846Z K 7 svn:log V 56 Add further ath_buf debugging, relevant for TX buffers. END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-14T04:13:48.519424Z K 7 svn:log V 52 Fix grammar mistakes. Submitted by: Benjamin Kaduk END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-14T07:56:25.178243Z K 7 svn:log V 1443 Add support to use the non read-and-copy interrupt register path. There appears to be a subtle race condition in the AR_ISR_RAC path where some of the secondary conditions would be not show up. This unfortunately includes TX events such as EOL. The reference driver includes an alternate path which instead uses the AR_ISR and AR_ISR_S{0,1,2,3,4,5} registers instead of AR_ISR_RAC and the shadow AR_ISR_S{0,1,2,3,4,5}_S registers. Here, the interrupts being handled are written back to the status register, clearing them. For interrupts caused by secondary registers, clear those bits instead of the relevant bit in AR_ISR. That way if an event occurs between the ISR_ISR_Sx read and write (clear), it won't be cleared; and it'll trigger another interrupt. This won't _entirely_ limit the TX hangs because of the existing txqactive() race going on between the interrupt handler and the TX process. I'll address that in a future commit. Note: I should also do this for the AR5212 series NICs as well as some of them may suffer from the same race. Finally - for now, never set the relevant HAL capability that I've introduced; I'll set it when I know which chips work and which don't. Finally finally - ath9k doesn't do this for the pre-ar9003 NICs but does for AR9300 v2.0 and later. ie, the AR9300 NIC interrupt code has this RAC capability check, and only enables it for AR9300 v2.0 MAC versions. Obtained from: Atheros, Linux ath9k END K 10 svn:author V 6 tuexen K 8 svn:date V 27 2011-09-14T08:15:21.785317Z K 7 svn:log V 190 Fix the handling of the flowlabel and DSCP value in the SCTP_PEER_ADDR_PARAMS socket option. Honor the net.inet6.ip6.auto_flowlabel sysctl setting. Approved by: re (bz) MFC after: 1 month. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-14T08:27:19.545184Z K 7 svn:log V 14 Update things END K 10 svn:author V 7 attilio K 8 svn:date V 27 2011-09-14T10:27:23.260396Z K 7 svn:log V 103 MFC r222951: Fix races in handling flags at detach time for FIBs. Sponsored by: Sandvine Incorporated END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-14T12:55:52.382262Z K 7 svn:log V 90 - Add some more comments because the code has changed recently - Some minor style changes END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-14T13:09:05.833862Z K 7 svn:log V 16 - Fix anchoring END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-14T13:27:24.487584Z K 7 svn:log V 97 - Remove the old regressiont tests, they are too specific to the old code and are not portable END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-14T14:37:54.709486Z K 7 svn:log V 28 Remove unused ds0 variable. END K 10 svn:author V 8 hselasky K 8 svn:date V 27 2011-09-14T15:16:53.693963Z K 7 svn:log V 335 Reduce USB memory usage during enumeration. We are allocating some kilobytes of extra memory during USB device enumeration. This does not change alot under FreeBSD, but makes sense for various embedded operating systems using the FreeBSD USB stack, which have less memory resources available. Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-14T15:47:18.909376Z K 7 svn:log V 103 Convert amrr/onoe over to use the new calling API. This is currently untested and doesn't support 11n. END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-14T18:07:18.941714Z K 7 svn:log V 98 - Add AT&T regex regression test and integrate it to our test suite. It has a BSD-like license. END K 10 svn:author V 6 tuexen K 8 svn:date V 27 2011-09-14T19:10:13.907914Z K 7 svn:log V 210 Ensure that 1-to-1 style SCTP sockets can only be connected once. Allow implicit setup also for 1-to-1 style sockets as described in the latest version of the socket API ID. Approved by: re MFC after: 1 month END K 10 svn:author V 8 brueffer K 8 svn:date V 27 2011-09-14T20:13:10.985556Z K 7 svn:log V 134 Minor spelling, wording and punctuation fixes in comments. PR: 155984 Submitted by: gcooper Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-14T21:08:02.145592Z K 7 svn:log V 138 - Update the old manual page to match better the specific features of TRE and drop those parts that are specific to the old regex code. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-15T03:59:56.692174Z K 7 svn:log V 138 Flip on autosleep support for Kite/Kiwi. It's only used in the status clearing of RX descriptors. I haven't yet figured out why that is. END K 10 svn:author V 5 dougb K 8 svn:date V 27 2011-09-15T04:07:40.159023Z K 7 svn:log V 36 Miscellaneous minor clarifications. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-15T04:16:03.319963Z K 7 svn:log V 65 Add some more local register definitions Obtained from: Atheros END K 10 svn:author V 2 ae K 8 svn:date V 27 2011-09-15T04:32:14.085105Z K 7 svn:log V 394 MFC r225445: Don't use the whole free space when resizing partition to a larger size on a disk with non zero stripesize (e.g. disks with 4k sector size)[1]. Also do not use automatic alignment when size is exactly specified, but an alignment is not. Use automatic alignment only for case when user omits both "-s" and "-a" options. Reported by: Mikael Fridh [1] END K 10 svn:author V 2 ae K 8 svn:date V 27 2011-09-15T04:34:59.848699Z K 7 svn:log V 160 MFC r223356 (by delphij): Mod the offset padding by alignment. Without this change we may pad too much when underlying GEOM object have a zero stripesize. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-15T05:05:31.432630Z K 7 svn:log V 115 Correct the AR_ISR_S5 logic. This still hasn't fixed the shared interrupt issue, however. Obtained from: Atheros END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-15T06:41:43.398486Z K 7 svn:log V 176 These aren't strictly RX_CLEAR hangs; but I'm adding them for now so I can see if they pop up. I'm seeing beacon hangs on Merlin when a second NIC is also busy doing traffic. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-15T06:42:06.321544Z K 7 svn:log V 47 Check for baseband hangs in the bstuck tasklet END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-15T08:42:06.083395Z K 7 svn:log V 359 Ensure that ta_pending doesn't overflow u_short by capping its value at USHRT_MAX. If it overflows before the taskqueue can run, the task will be re-added to the taskqueue and cause a loop in the task list. Reported by: Arnaud Lacombe Submitted by: Ryan Stone Reviewed by: jhb Approved by: re (kib) MFC after: 1 day END K 10 svn:author V 6 tuexen K 8 svn:date V 27 2011-09-15T08:49:54.191060Z K 7 svn:log V 130 Make sure that SCTP rejects broadcast, multicast and wildcard addresses as remote addresses. Approved by: re MFC after: 1 month. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-15T09:06:28.043310Z K 7 svn:log V 214 Bring over a HAL function from the reference code which calculates the amount of time spent in TX/RX/channel busy. This is just for debugging; it shouldn't be merged in this form to -HEAD. Obtained from: Atheros END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-15T09:22:46.173613Z K 7 svn:log V 392 Fix the interrupt mitigation stuff, now that I know what's going on. * If the RX mitigation interrupts fire, set HAL_INT_RX and clear the relevant underlying interrupt(s) as well. * If the TX mitigation interrupts fire, set HAL_INT_TX * If any of the TX status bits (ok, err, desc, eol) are set, update the TX bitmask, but don't set HAL_INT_TX if TX interrupt mitigation is enabled. END K 10 svn:author V 5 dougb K 8 svn:date V 27 2011-09-15T09:22:49.881308Z K 7 svn:log V 109 Correct the RFC number for the description of IPv6 privacy addressing Reviewed by: bz Approved by: re (kib) END K 10 svn:author V 3 kib K 8 svn:date V 27 2011-09-15T09:53:04.408316Z K 7 svn:log V 272 Microoptimize the return path for the fast syscalls on amd64. Arrange the code to have the fall-through path to follow the likely target. Do not use intermediate register to reload user %rsp. Proposed by: alc Reviewed by: alc, jhb Approved by: re (bz) MFC after: 2 weeks END K 10 svn:author V 3 kib K 8 svn:date V 27 2011-09-15T09:54:07.324916Z K 7 svn:log V 124 Put amd64_syscall() prototype in md_var.h. Requested by: jhb Reviewed by: alc, jhb Approved by: re (bz) MFC after: 2 weeks END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-15T10:35:38.769197Z K 7 svn:log V 82 MFC r209030: In dtrace lexer, do not unput token if it is EOF. On behalf of: kan END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-15T10:39:44.644234Z K 7 svn:log V 121 MFC r209305,209358: Do not allow EOF token to be put back into input buffer. PR: kern/159064 On behalf of: kan, marcel END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-15T10:42:55.310688Z K 7 svn:log V 121 MFC r209030,209305,209358: In dtrace lexer, do not unput token if it is EOF. PR: kern/159064 On behalf of: kan, marcel END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-15T10:51:31.796139Z K 7 svn:log V 510 MFC r213736: generic_stop_cpus: prevent parallel execution Addendum to the original commit message: This is mostly useful to protect kdb_trap() from concurrent entry as it doesn't have any protection like e.g. panic(9) has. Even if kdb_trap() had that protection there would still be a need for protection between concurrect panic() and kdb_trap() entries. That protection could be implemented externally, but doing it in generic_stop_cpus() seems to be OK as well (stopping CPUs on panic is in the works). END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-15T11:17:07.484651Z K 7 svn:log V 73 MFC r223530: generic_stop_cpus: pull timeout logic from under DIAGNOSTIC END K 10 svn:author V 3 kib K 8 svn:date V 27 2011-09-15T11:50:09.264577Z K 7 svn:log V 237 Use the proper dynamic tls block to calculate the tls variable address in case tls data generation was updated. PR: misc/160721 Submitted by: "Thinker K.F. Li" Tested by: flo Approved by: re (bz) MFC after: 1 week END K 10 svn:author V 8 lstewart K 8 svn:date V 27 2011-09-15T12:15:36.586012Z K 7 svn:log V 295 Rename the cc.4 and cc.9 modular congestion control related man pages to mod_cc.4 and mod_cc.9 respectively to avoid any possible confusion with the cc.1 gcc man page. Update references to these man pages where required. Requested by: Grenville Armitage Approved by: re (kib) MFC after: 3 days END K 10 svn:author V 6 tuexen K 8 svn:date V 27 2011-09-15T12:20:52.975625Z K 7 svn:log V 137 Fix a typo introduced in http://svn.freebsd.org/changeset/base/225571 Reported by Ilya A. Arkhipov. Approved by: re MFC after: 1 month. END K 10 svn:author V 7 attilio K 8 svn:date V 27 2011-09-15T12:27:26.448785Z K 7 svn:log V 148 MFC r225177,225181: Introduce and use seldrain() function for dealing with fast selrecord/selinfo destruction. Sponsored by: Sandvine Incorporated END K 10 svn:author V 2 ae K 8 svn:date V 27 2011-09-15T12:28:17.075987Z K 7 svn:log V 291 Add IPv6 support to the ng_ipfw(4) [1]. Also add ifdefs to be able build it with and without INET/INET6 support. Submitted by: Alexander V. Chernikov [1] Tested by: Alexander V. Chernikov [1] Approved by: re (bz) MFC after: 2 weeks END K 10 svn:author V 2 jh K 8 svn:date V 27 2011-09-15T13:32:43.035512Z K 7 svn:log V 175 Expose "log" in the default devfs rules. /etc/rc.d/jail creates /dev/log as a symbolic link. PR: conf/160711 Submitted by: Jase Thew Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-15T15:09:45.400127Z K 7 svn:log V 27 - Merge bugfixes from grep END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-15T15:12:18.250798Z K 7 svn:log V 18 - Merge from grep END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-15T15:18:57.839101Z K 7 svn:log V 94 - Simplify REG_PEND code and add support for the same in wchar-compliant regcomp() variants END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-15T15:28:41.397050Z K 7 svn:log V 71 - Check nmatch not only REG_NOSUB if we can store submatches in pmatch END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-15T15:54:37.188863Z K 7 svn:log V 135 - Refactor code to only call fast matcher and heuristic code if input is not STR_USER. This makes reguexec() work again like before. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-15T17:11:03.478031Z K 7 svn:log V 62 Merge in beacon related fixes (mostly STA related) from HEAD. END K 10 svn:author V 7 yongari K 8 svn:date V 27 2011-09-15T17:20:20.325988Z K 7 svn:log V 833 MFC r225440: vge(4) hardwares poll media status and generates an interrupt whenever the link state is changed. Using software based polling for media status tracking is known to cause MII access failure under certain conditions once link is established so vge(4) used to rely on link status change interrupt. However DEVICE_POLLING completely disables generation of all kind of interrupts on vge(4) such that this resulted in not detecting link state change event. This means vge(4) does not correctly detect established/lost link with DEVICE_POLLING. Losing the interrupt made vge(4) not to send any packets to peer since vge(4) does not try to send any packets when there is no established link. Work around the issue by generating link state change interrupt with DEVICE_POLLING. PR: kern/160442 END K 10 svn:author V 7 yongari K 8 svn:date V 27 2011-09-15T17:22:38.862078Z K 7 svn:log V 833 MFC r225440: vge(4) hardwares poll media status and generates an interrupt whenever the link state is changed. Using software based polling for media status tracking is known to cause MII access failure under certain conditions once link is established so vge(4) used to rely on link status change interrupt. However DEVICE_POLLING completely disables generation of all kind of interrupts on vge(4) such that this resulted in not detecting link state change event. This means vge(4) does not correctly detect established/lost link with DEVICE_POLLING. Losing the interrupt made vge(4) not to send any packets to peer since vge(4) does not try to send any packets when there is no established link. Work around the issue by generating link state change interrupt with DEVICE_POLLING. PR: kern/160442 END K 10 svn:author V 8 dwmalone K 8 svn:date V 27 2011-09-15T19:52:43.658897Z K 7 svn:log V 177 It seems when I added code for affine unit changes to units, I forgot to tell the man page we could now convert Celsius to Fahrenheit. Approved by: re (kib) MFC after: 1 month END K 10 svn:author V 3 mdf K 8 svn:date V 27 2011-09-15T20:21:35.880841Z K 7 svn:log V 130 Welcome Gleb Kurtsou (gleb at FreeBSD dot org). Gleb has worked on the ino64 GSoC project among other things. Approved by: core END K 10 svn:author V 6 grehan K 8 svn:date V 27 2011-09-15T22:14:35.978151Z K 7 svn:log V 79 IFC @ r225592 sys/dev/bvm/bvm_console.c - move up to the new alt-break order. END K 10 svn:author V 3 des K 8 svn:date V 27 2011-09-15T22:50:31.879601Z K 7 svn:log V 754 When resuming an HTTP download, we failed to verify that the range returned by the server matched what we requested, and blindly appended what we received to what we already had. This could go two ways: if the delivered offset was higher than expected, the local file would contain duplicate data, while if it was lower than expected, there would be data missing from the middle of the file. Furthermore, if the transfer was interrupted again, each subsequent attempt would compound the error. Fix the first problem by restarting the transfer from scratch if there is a gap, and the second by explicitly seeking to the correct location in the local file so as to overwrite any duplicated data. PR: bin/117277 Approved by: re (kib) MFC after: 3 weeks END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-16T02:28:55.488051Z K 7 svn:log V 294 Merge r225570 from HEAD - Ensure that ta_pending doesn't overflow u_short by capping its value at USHRT_MAX. If it overflows before the taskqueue can run, the task will be re-added to the taskqueue and cause a loop in the task list. Submitted by: rstone@ Reviewed by: jhb@ MFC after: 1 day END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-16T04:05:58.762823Z K 7 svn:log V 932 The ar5212InitState() function sets up AH5212(ah)->ah_beaconInterval to be 100 TU's by default and I'm guessing this once matched the default beacon interval as specified in the ath driver. This is then used in the CAB queue setup code in AR5212/AR5416 to setup the CAB queue ready time value. This is how long the CAB queue gets to transmit frames before it's halted, so other lower priority queues (ie, the data queues) get a chance to TX. But now that staggered beacons are enabled, the CAB queue readytime value will be shorter. The beacon interval timer is now 25 TU's rater than 100 TU's. So specifying a CAB readytime greater than the TBTT beacon timer interval is likely a bad idea. This hasn't resolved the issues I've seen with CAB queue traffic not draining, nor quietened missing beacon TX. But it certainly looks a lot more correct. Some further investigation into exactly how the cabq should be setup is warranted. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-16T04:09:53.930095Z K 7 svn:log V 160 Update the beacon interval value here; hopefully the queue setup will occur -after- the beacon interval has been set. TODO: verify this is actually the case.. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-16T04:26:40.515632Z K 7 svn:log V 72 No need to gate the CABQ if the queue is empty. Obtained from: Atheros END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-16T04:42:05.176474Z K 7 svn:log V 1652 Fix the default cabq configuration time to not be "too long". The -HEAD code would configure a cabq time of 100 TU minus the various beacon setup intervals. When doing staggered beacons, this would result in CABQ readytime being 4 * the beacon interval timer, which is 1/ATH_BCBUF (4) the default beacon interval (100 TU.) This didn't seem to annoy the AR9160 MAC but it definitely annoys the AR9220/AR9280 MAC. I then fixed it to use the beacon time configured through a (hopefully earlier!) call to the HAL beacon timer setup code. This reduced the stuck beacon and cabq stall occurances quite a bit, but it didn't eliminate them. Having a CABQ time that's almost the same as the beacon interval seems to annoy AR9280 (Merlin) and later MACs. So, I decided to twiddle things a bit and make it 70% of the beacon interval. Both linux ath9k and the Atheros reference code default to 80% of the beacon interval. This hasn't stopped miss beacons (which could be because the air is "busy", which it tends to be in my apartment..) or eliminated the cabq TX issues (which should still be addressed, as there's no way a few broadcast frames here and there shouldn't make it out when the air is completely free for TX) but it does seem to have (initially) eliminated the stuck beacon conditions on my AR9220 NIC. I'll have to run this for a few hours to see what the story is and I likely should investigate why cabq traffic can't be TXed, but I'm happy to report these last few commits have had a positive effect. TODO: fix this in AR5212 too, just to be correct. TODO: actually set a non-0 readytime value when setting up the CABQ in the ath driver. END K 10 svn:author V 7 delphij K 8 svn:date V 27 2011-09-16T05:45:13.806014Z K 7 svn:log V 419 MFC r225338: Fix the check in dircheck() on namlen. The value of namlen is copied from on-disk d_namlen, which is a 8-bit unsigned integer which can never exceed MAXNAMLEN (255) so the test is always true. Moreover, UFS does not allow d_namelen being zero. Change namlen from u_int to u_int8_t, and replace the unneeded test with a useful test. PR: bin/160339 Submitted by: Eugene Grosbein END K 10 svn:author V 7 delphij K 8 svn:date V 27 2011-09-16T05:57:01.682657Z K 7 svn:log V 186 MFC r225340: Fix a bug in ichwd(4) which prevents it from being enabled if the new timeout is the same timeout. Submitted by: Dmitrij Tejblum PR: kern/139604 END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-16T06:33:12.334090Z K 7 svn:log V 278 Tidy up the CAB queue defaults; do the same for AR5212. * add a check to the AR5416 CABQ setup to ensure a minimum cabq time; that way a negative calculated cabq time doesn't annoy things; * Port the AR5416 CABQ setup stuff to AR5212. TODO: still do this in the driver.. END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-16T08:22:48.197568Z K 7 svn:log V 90 zfstest: rename to zfsboottest and move to tools Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 3 avg K 8 svn:date V 27 2011-09-16T08:24:31.241988Z K 7 svn:log V 270 zfsboottest: some additional enhancements - redirect diagnostics printfs in the boot code to stderr - do not read trailing garbage from a trailing block of a file Also add my copyright to the file after making so many changes. Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 7 pluknet K 8 svn:date V 27 2011-09-16T09:09:58.971218Z K 7 svn:log V 246 Print the package name on deletion errors. It appears this was already done in NetBSD a decade ago, hence I just reuse the change (except our code is bad styled). PR: bin/160516 Approved by: portmgr Approved by: re (kib) Obtained from: NetBSD END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-16T09:15:34.850505Z K 7 svn:log V 18 Merge from -HEAD. END K 10 svn:author V 10 nwhitehorn K 8 svn:date V 27 2011-09-16T09:37:13.348294Z K 7 svn:log V 141 Fix WPA setup broken by changes to shell argument passing inside the installer network configuration. PR: bin/159558 Approved by: re (kib) END K 10 svn:author V 10 nwhitehorn K 8 svn:date V 27 2011-09-16T09:38:01.406537Z K 7 svn:log V 117 Fix a typo introduced in r219892 that prevented file system options from being set correctly. Approved by: re (kib) END K 10 svn:author V 3 des K 8 svn:date V 27 2011-09-16T11:14:10.741468Z K 7 svn:log V 117 Remove the svn:keywords property and restore the historical $FreeBSD$ tag. Approved by: re (kib) MFC after: 3 weeks END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-16T12:12:26.049789Z K 7 svn:log V 37 - Include missing header when needed END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-16T12:13:03.965828Z K 7 svn:log V 64 - Extract duplicated macro into a function to avoid duplication END K 10 svn:author V 5 kmacy K 8 svn:date V 27 2011-09-16T13:58:51.513683Z K 7 svn:log V 558 In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz) END K 10 svn:author V 5 kmacy K 8 svn:date V 27 2011-09-16T14:04:14.005813Z K 7 svn:log V 84 Auto-generated code from sys_ prefixing makesyscalls.sh change Approved by: re(bz) END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T17:39:39.694365Z K 7 svn:log V 152 MFC 223936,224205: Allow per-CPU statistics to be toggled at runtime via the 'P' key. While here, make -P a toggle similar to other options such as -I. END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T17:41:56.732235Z K 7 svn:log V 148 MFC 224062: Revert 130163 and let top use KERN_PROC_PROC when individual threads are not displayed. The KERN_PROC_PROC sysctl was fixed in 188764. END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T17:44:33.979670Z K 7 svn:log V 108 MFC 224039: Use MFI_STAT_* constants for cmd_status field values rather than magic numbers in a few places. END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T17:47:12.668511Z K 7 svn:log V 115 MFC 224041: Support controllers whose option ROM is disabled in the BIOS by kicking the firmware so that it boots. END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T18:18:38.389311Z K 7 svn:log V 185 MFC 224228: Note that the "kernel" variable in loader.conf is the name of a directory containing a kernel under /boot and that it's default value is "kernel" not "/boot/kernel/kernel". END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T18:23:39.711370Z K 7 svn:log V 170 MFC 224733: Allow the NFS client to use a max file size larger than 1TB for v3 mounts. It now allows files up to OFF_MAX subject to whatever limit the server advertises. END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T18:26:40.755467Z K 7 svn:log V 495 MFC 225079: Add two new macros DRIVER_MODULE_ORDERED() and EARLY_DRIVER_MODULE_ORDERED() that allow a module to use an order other than the default of SI_ORDER_MIDDLE when registering a driver. This can be useful for drivers in a kld that contain multiple new-bus drivers as part of one logical device driver. A typical case would be to use SI_ORDER_LAST for the "main" driver to ensure that any other "helper" drivers are registered and available before the "main" driver attempts to attach. END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T18:36:43.519510Z K 7 svn:log V 118 MFC 225194: Make NKPT a kernel option on amd64 so that it can be set to a non-default value from kernel config files. END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T18:41:19.545603Z K 7 svn:log V 178 MFC 225201: Enable the puc(4) driver on amd64 and i386 in GENERIC. This allows devices supported by puc(4) to work "out of the box" since puc.ko does not work "out of the box". END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T19:29:11.905694Z K 7 svn:log V 162 MFC 224942: Fix a regression where a rule containing a source port option after a destination IP would incorrectly display the source port as a destination port. END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T19:52:49.134157Z K 7 svn:log V 212 MFC 225017: Walk the zombproc list as well as the allproc list when enumerating threads and processes in a kernel image. This allows examination of threads that have exited or are in the late stages of exiting. END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-16T19:53:56.074779Z K 7 svn:log V 189 MFC 225200: - Replace references to sio(4) with uart(4) instead. - We no longer use the same data structure in as NetBSD in pucdata.c. - ppc(4) has had a puc(4) attachment for a while now. END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-16T22:38:57.416889Z K 7 svn:log V 56 - Factor out duplicated macro into a common header file END K 10 svn:author V 2 cs K 8 svn:date V 27 2011-09-16T22:57:50.094324Z K 7 svn:log V 70 Add myself and connect me with my mentors Approved by: re@, glarkin@ END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-17T04:31:43.451940Z K 7 svn:log V 132 Flip the AIFS default for beaconq to be in line with the ath9k/reference driver. TODO: is this valid for the ath5k chips, perhaps? END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-17T05:35:59.860419Z K 7 svn:log V 37 - Portability fix for format strings END K 10 svn:author V 6 tuexen K 8 svn:date V 27 2011-09-17T08:50:29.290470Z K 7 svn:log V 152 Fix the enabling/disabling of Heartbeats and path MTU discovery when using the SCTP_PEER_ADDR_PARAMS socket option. Approved by: re MFC after: 1 month. END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-17T08:54:23.514464Z K 7 svn:log V 90 - Reword and fix some parts of the manual Submitted by: Ben Kaduk END K 10 svn:author V 10 nwhitehorn K 8 svn:date V 27 2011-09-17T09:25:45.258686Z K 7 svn:log V 242 Inquire about terminal type if starting from a serial console and do not configure the host's networking if netbooted [1]. Also fix FTP installations behind some firewalls [2]. PR: bin/159583 [2] Reported by: stas [1] Approved by: re (kib) END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-17T10:24:19.487998Z K 7 svn:log V 11 - Fix typo END K 10 svn:author V 7 attilio K 8 svn:date V 27 2011-09-17T11:08:40.008103Z K 7 svn:log V 4 MFC END K 10 svn:author V 6 rstone K 8 svn:date V 27 2011-09-17T13:48:09.754170Z K 7 svn:log V 286 Clear transmit checksum offload context state upon lem(4) interface initialization. Prior to this change packets may be transmitted with an incorrect checksum. Em(4) already has an equivalent change in r213234. Obtained From: Sandvine MFC After: 1 week Approved by: re (bz) END K 10 svn:author V 5 trasz K 8 svn:date V 27 2011-09-17T19:55:32.936415Z K 7 svn:log V 366 Fix long-standing thinko regarding maxproc accounting. Basically, we were accounting the newly created process to its parent instead of the child itself. This caused problems later, when the child changed its credentials - the per-uid, per-jail etc counters were not properly updated, because the maxproc counter in the child process was 0. Approved by: re (kib) END K 10 svn:author V 5 trasz K 8 svn:date V 27 2011-09-17T20:48:49.272399Z K 7 svn:log V 140 Fix error handling bug that would prevent MAC structures from getting freed properly if resource limit got exceeded. Approved by: re (kib) END K 10 svn:author V 7 delphij K 8 svn:date V 27 2011-09-17T22:17:31.545013Z K 7 svn:log V 232 Change the example of using smbfs in fstab to a password-less one, which is less likely to block a remote system from boot. Submitted by: Garrett Cooper PR: doc/160775 MFC after: 1 month Approved by: re (kib) END K 10 svn:author V 7 delphij K 8 svn:date V 27 2011-09-17T22:53:05.846905Z K 7 svn:log V 94 Bump .Dd date for previous revision. Approved by: re (follow up commit fixing trivial issue) END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-18T07:20:59.744480Z K 7 svn:log V 1350 Begin an overhaul of the TXQ locking to avoid recursive locking. The linux and atheros reference driver keep the hardware TXQ locked for the duration of both the TXQ completion processing and TXQ drain. This unfortunately means the TXQ lock is held during completion callbacks and during ieee80211_free_node(). When the last node reference is removed, net80211 will free the node - which calls back into the driver and in this code, it flushes the node state and removes it from various lists. Unfortunately access to these lists is locked by the TXQ lock and trying to grab it whilst it's already being held causes a panic. The solution is to go back to how FreeBSD did the locking - hold the lock whilst doing TXQ manipulation (add/remove/state changes) and then reacquire it whilst fiddling with the per-TID state. The lock isn't to be held across TX buffer completion. This commit partially implements this: * the non-aggregation pathway doesn't hold the lock across the completion handler * the aggregation pathway for single packets doesn't hold the lock across the completion handler * the aggregation pathway for aggregate frames now delays calling the completion handler until the TID state (BAW, etc) has been updated; it's done at the end of the function. The error and cleanup pathways will be fixed in a subsequent commit. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-18T08:26:57.720252Z K 7 svn:log V 296 Part 2 of the locking/completion overhaul. Update the rest of the flush, error and cleanup functions to delay calling the completion functions until the TXQ lock has been released. The TXQ locking in some instances is quite inefficient and should be tidied up before this is merged into -HEAD. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-18T09:05:42.789908Z K 7 svn:log V 680 Enable printf's where a BAR frame _would_ be TXed. I'm doing this so I (and testers) can keep track of when BAR's should be TXed. This way if traffic pauses are seen that coincide with this BAR message, we can (hopefully!) assume that a BAR frame exchange would've flushed things correctly. This also will let me implement/debug the "correct" sending of BAR frames - ie, sending them when the TID has been emptied rather than on each failed software retry attempt. That could get slightly spammy. There's a few places where BAR frames should likely be TX'ed, specifically when a TID flush is done (and the node isn't being deleted.) But that'll come in the next phase of work. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-18T09:42:21.356095Z K 7 svn:log V 146 Move the lock around so it happens after the rate control update is done. (That way the ath_node lock and txq lock aren't held at the same time.) END K 10 svn:author V 3 kib K 8 svn:date V 27 2011-09-18T11:07:51.296939Z K 7 svn:log V 42 Welcome Robert Millan. Approved by: core END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-18T13:05:39.183978Z K 7 svn:log V 78 Rename the RX clear hangs so I get a better idea of who/what is causing them. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-18T13:28:56.424591Z K 7 svn:log V 39 Update the comment to reflect reality. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-18T14:17:15.825224Z K 7 svn:log V 371 Some TID sched/unsched changes to try and better balance per-node traffic. The reference driver does a basic FIFO implementation for scheduling traffic from TIDs to the hardware queue. This is an implementation of that, inspired by Linux ath9k. Whilst I'm at it, eliminate a unneeded parameter from the tid sched/unsched functions. Obtained from: Linux ath9k, Atheros END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-18T15:11:54.040529Z K 7 svn:log V 237 Make sure a paused queue gets some frames re-scheduled to the hardware once it's unpaused or it'll hang until the next TX'ed frame occurs. Pointy hat to: adrian, for removing this call a while back and never remembering to restore it. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-19T00:46:49.656655Z K 7 svn:log V 253 Revert the FIFO TX scheduling part of the previous commit. There's a bug somewhere in the code which causes TX latency to jump dramatically and this indicates some scheduling bugs. I'll squeeze out those bugs and then re-commit the FIFO TX scheduling. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-19T04:06:56.372091Z K 7 svn:log V 38 Flip on cleanup when disabling AMPDU. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-19T04:08:52.539318Z K 7 svn:log V 157 Add a comment - there's no need to update the BAW here. It's likely I'll have to fiddle with the BAW flags though, or debugging messages will be triggered. END K 10 svn:author V 8 jpaetzel K 8 svn:date V 27 2011-09-19T05:12:53.405449Z K 7 svn:log V 494 Fix a logic bug in pc-sysinstall creating partitions. Improve exit when an error occurs. Fix parsing to grab values which contain extra '=' signs. Fix a bug setting the timezone properly. Fix a usage bug when setting up with gmirror. Allow a uzip file from local media to be used. Allow specifying flags for "newfs" when using UFS as the file system. Run custom commands after doing final cleanup / fstab generation and such. Also fix using relative path for config file. Approved by: re (bz) END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-19T08:01:21.373190Z K 7 svn:log V 1123 First cut at enabling the TX of BAR frams. Currently ieee80211_send_bar() calls back into the driver via ic->ic_raw_xmit(). The driver has some special case code to directly dispatch BAR frames for this exact purpose. This means that the hardware TXQ lock is needed, so BAR frame TX must occur from _outside_ the TXQ lock. * Unlock the TXQ before attempting to send a BAR frame * Take a note of the txseq whilst inside the TXQ lock - although other contexts may allocate TX sequence numbers (and I'm about to stick that inside the TXQ lock too just to be safe), all frames from this point on have not yet attempted to be transmitted. So the TX sequence number at this point is fine as a left edge for the BAW. This has only received light testing due to some bugs I've introduced which make a fixed unicast rate no longer work (the multi-rate retry schedule is incorrectly being set.) This means that although I can trigger frame failure, it isn't predictable or guaranteed. I'll fix that in a subsequent commit and then properly test the aggregate and non-aggregate BAR pathways whilst traffic is ongoing. END K 10 svn:author V 8 hselasky K 8 svn:date V 27 2011-09-19T08:52:06.958587Z K 7 svn:log V 92 Implement missing USB debug information functions. Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-19T09:52:03.541099Z K 7 svn:log V 58 Un-break the handling of multi-rate retry in the TX path. END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-19T10:28:59.550857Z K 7 svn:log V 74 - Treat literal patterns really literal Submitted by: aakuusta@gmail.com END K 10 svn:author V 7 attilio K 8 svn:date V 27 2011-09-19T10:58:30.734482Z K 7 svn:log V 263 #PROCHOT assertion is sticky after reading the MSR (accordingly with Intel manuals) it must be cleared by writing a 0. Fix that. Sponsored by: Sandvine Incorporated Reported by: rstone Reviewed by: delphij, emaste, rstone Approved by: re (kib) MFC after: 1 week END K 10 svn:author V 7 attilio K 8 svn:date V 27 2011-09-19T11:08:31.062425Z K 7 svn:log V 63 MFC r225448: Improve busy buffers diagnostic on shutdown path. END K 10 svn:author V 6 adrian K 8 svn:date V 27 2011-09-19T11:49:05.922269Z K 7 svn:log V 221 Add some further debugging to BAW add/remove. The current TX hangs I've seen with BAR TX are because I'm completely doing the wrong thing in choosing a new left-edge BAW. I'll look at fixing that in subsequent commits. END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-19T13:39:40.027062Z K 7 svn:log V 36 - Remove unused macro - Sort macros END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-19T13:41:12.142097Z K 7 svn:log V 11 - Fix typo END K 10 svn:author V 5 gabor K 8 svn:date V 27 2011-09-19T13:55:19.416955Z K 7 svn:log V 39 - Fix regression introduced in r225661 END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-19T15:31:12.768525Z K 7 svn:log V 183 Copy ip6po_minmtu and ip6po_prefer_tempaddr in ip6_copypktopts(). This fixes inconsistency when options are specified by both setsockopt() and ancillary data types. PR: kern/158307 END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-19T15:55:53.200362Z K 7 svn:log V 149 Fix the script order to run rc.d/bridge after the initial network interface configuration and before running network daemons. Approved by: re (kib) END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-19T15:56:50.287403Z K 7 svn:log V 96 Use resolvconf(8) to create /etc/resolv.conf, not directly overwrite it. Approved by: re (kib) END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-19T16:00:55.901259Z K 7 svn:log V 300 - Document $ipv6_cpe_wanif. - Emphasize $ipv6_enable and $ipv6_prefer are deprecated. - Add more detail descriptions about $ipv6_activate_all_interfaces. - Add some more examples of $ifconfig_IF_ipv6. - rtsold(8) and rtadvd(8) can be used even when ipv6_gateway_enable=NO now. Approved by: re (kib) END K 10 svn:author V 3 hrs K 8 svn:date V 27 2011-09-19T16:01:53.755060Z K 7 svn:log V 162 - Document inet6 no_radr flag. - Add descriptions of sysctl(8) variables which can control the default configuration of the inet6 flags. Approved by: re (kib) END K 10 svn:author V 3 jhb K 8 svn:date V 27 2011-09-19T16:25:37.020472Z K 7 svn:log V 226 MFC 225331: Move the logic to parse volume cache commands out into a separate function and use a loop so that multiple cache commands can be strung together on the command line into a single update to the volume's properties. END