IABSD.fr/src/sys
Branch :
Log
| Author | Commit | Date | CI | Message |
|---|---|---|---|---|
 |
bf77f779 | 2025-05-24 06:50:02 | sync | |
|
75937829 | 2025-05-24 06:49:16 | In the old gprof profiling subsystem, the simplistic profil() syscall told the kernel about the sample buffer, and then the normal exit-time _mcleanup() would finalize the buffer, open()'ed a file and write out the details. This file opening has become increasingly impossible because of our privsep / privdrop, chroot, setresuid uid-dropping, pledge, unveil, and other efforts. So people stopped using gprof. Programs which needed profiling needed substantial mitigation removal changes to put them under test. In the new gprof profiling subsystem, profil() is enhanced to provide more buffer information to the kernel, as well as better instructions for where the file should be written. At normal exit-time mcleanup(), after finalizing the buffer, the process simply terminates via _exit(2), and the kernel does all the opening and writing in a safe way. The file is now deposited into the starting directory, or into PROFDIR as determined at program start, with safety issues taken into consideration. Additional side effecs: - The monstartup(3) interface will go away, it cannot be supported. - profil() is now allowed in pledge "stdio", access is decided by the -pg ELF note - gmon.out is renamed to gmon.progname.pid.out, due to pervasive multi-process privsep practices prompted originally by job and claudio for use in rpki-client and bgpd advice from kettenis and claudio | |
 |
173f4495 | 2025-05-24 00:19:09 | timeout_add_ticks doesnt have to do rounding now it's callers do it. timeout_add_ticks turned short (less than 1 tick but greater than 0) timeouts into 1 tick timeouts. timeout_add_nsec/usec/msec round themselves up now, so this isn't necessary anymore. | |
|
86ddee2d | 2025-05-24 00:11:08 | have timeout_add_nsec/usec/msec wait at least their specified time. timeouts run off kernel clock ticks which are longer than nsec/usec/msec intervals, so simply dividing these values by how long a tick is will truncate the desired times. to mitigate this, round up the value before the divide. rounding up can lead to integer overflow, so detect this and clamp the value to the max sleep time. while here get timeout_add_sec to check for a negative time like timeout_add does to the specified tick interval. discussed with many help with the int overflow stuff from djm@ ok sashan@ | |
|
a86a4190 | 2025-05-23 23:56:14 | deprecate timeout_add_tv now that nothing uses it anymore. i feel like it was a bit of a footgun because timeouts had to assume that the timeval passed is valid. the api only returns whether the timeout was added or already pending, it's not clear what it should do if the timeval is invalid. ok mpi@ bluhm@ | |
|
b00facf8 | 2025-05-23 23:41:46 | replace timeout_add_tv with timeout_add_nsec via TIMEVAL_TO_NSEC. this is the last use of timeout_add_tv, which id like to deprecate. ok mp@ bluhm@ | |
|
403e61e5 | 2025-05-23 23:39:30 | replace timeout_add_tv with timeout_add_usec this changes the calculation of the interval slightly, but in practice it should not make a difference. this is a step toward deprecating timeout_add_tv. ok mpi@ bluhm@ | |
 |
c75d946f | 2025-05-23 11:50:21 | remove unused ice(4) softc fields inherited from FreeBSD iflib if_softc_ctx Leaving TSO-related ones in place for now, since TSO is still a todo item. | |
 |
09743abc | 2025-05-23 09:56:26 | Set an RSS key and lookup table so traffic is properly distributed across the queues, and copy the RSS hash into received packets as the flow ID. from Yuichiro NAITO, lightly modified | |
|
f3256d04 | 2025-05-23 09:16:14 | enable RSS in ice(4), and enable Tx/Rx across multiple queues Port ice_upd_prof_hw() and related functions from FreeBSD which were still missing from the RSS init code path. With 9k jumbo frames I am now seeing about 10Gbit/s throughput in multi-threaded iperf benchmarks against FreeBSD ice(4). | |
|
19fb9659 | 2025-05-23 09:11:06 | remove unused and redundant isc_nrxqsets and isc_ntxqsets variables from ice(4) | |
|
dc13af7a | 2025-05-23 08:57:49 | implement ice(4) Tx checksum offloading test + ok bluhm@ | |
 |
cad777a1 | 2025-05-23 03:13:33 | stacktrace_save_utrace() should be more robust Currently the function assumes the frame pointer address which comes from userland is always valid frame pointer address. Unfortunately if the frame address gets corrupted such it fits a kernel memory address range, then stacktrace_save_utrace() function just dereferences the address causing a kernel crash. The fix makes sure that as soon as stack unwinding crosses kernel<->userland boundary it never attempts to interpret the frame address to be coming from kernel. So after crossing the boundary all frame pointer addresses are passed to copyin(9f) to obtain a next frame. copyin(9f) then may fail with error indicating the stack is corrupted and stack unwinding process should stop. The fix was discussed at tech@. The idea to use two loops comes from tedu@ There is also a todo note from claudio@: skip iteration over kernel frames just find the first userland's frame and start there. OK tedu@ | |
 |
449dd8e0 | 2025-05-23 03:06:09 | Add preliminary support for RTL8157. Tested: RTL8153 (0x5c20/0x5c30): stsp | |
|
1a842094 | 2025-05-23 03:05:05 | sync | |
|
191dd14a | 2025-05-23 03:04:43 | Add USB product ID for RTL8157 | |
 |
1b0dae99 | 2025-05-23 02:50:08 | handle in_s4 moving to pm notifier | |
|
5a206aca | 2025-05-23 02:48:47 | drm/amdgpu: fix pm notifier handling From Alex Deucher 25e07c8403f4daad35cffc18d96e32a80a2a3222 in linux-6.12.y/6.12.30 4aaffc85751da5722e858e4333e8cf0aa4b6c78f in mainline linux | |
|
5c0b549e | 2025-05-23 02:47:09 | drm/fbdev-dma: Support struct drm_driver.fbdev_probe From Thomas Zimmermann 25998ed87f0c8772bd5faf8ee92d831f1c69e1d0 in linux-6.12.y/6.12.30 8998eedda2539d2528cfebdc7c17eed0ad35b714 in mainline linux | |
|
f621b50a | 2025-05-23 02:44:18 | dma-buf: insert memory barrier before updating num_fences From Hyejeong Choi fe1bebd0edb22e3536cbc920ec713331d1367ad4 in linux-6.12.y/6.12.30 72c7d62583ebce7baeb61acce6057c361f73be4a in mainline linux | |
|
6f46f340 | 2025-05-23 02:42:45 | drm/amd/display: Avoid flooding unnecessary info messages From Wayne Lin 64a33088adf2b4b23f34566d18ac73e8b4f64243 in linux-6.12.y/6.12.30 d33724ffb743d3d2698bd969e29253ae0cff9739 in mainline linux | |
|
bfa243c3 | 2025-05-23 02:41:09 | drm/amd/display: Correct the reply value when AUX write incomplete From Wayne Lin 2d125a1f0d61bf224ed7fe6ce2f4f7d4f37b4c61 in linux-6.12.y/6.12.30 d433981385c62c72080e26f1c00a961d18b233be in mainline linux | |
|
98b3332a | 2025-05-23 02:39:20 | drm/amdgpu: csa unmap use uninterruptible lock From Philip Yang 8d71c3231b33e24a911b8f2d8c3a17ee40aa32d5 in linux-6.12.y/6.12.30 a0fa7873f2f869087b1e7793f7fac3713a1e3afe in mainline linux | |
|
ecabef8d | 2025-05-23 02:37:58 | drm/amdgpu: fix incorrect MALL size for GFX1151 From Tim Huang e54f20c5a17e8dff4bb54f4f88ec098dc905ee41 in linux-6.12.y/6.12.30 2d73b0845ab3963856e857b810600e5594bc29f4 in mainline linux | |
|
5133bdf5 | 2025-05-23 02:36:37 | Revert "drm/amd/display: Hardware cursor changes color when switched to software cursor" From Melissa Wen c75f3f9cbfb7a84254285c587840afe4494d6992 in linux-6.12.y/6.12.30 fe14c0f096f58d2569e587e9f4b05d772272bbb4 in mainline linux | |
|
4d9edfda | 2025-05-23 02:35:08 | Revert "drm/amd: Stop evicting resources on APUs in suspend" From Alex Deucher 5d9d62323e1893e6f429fad7da19a652088cac21 in linux-6.12.y/6.12.30 d0ce1aaa8531a4a4707711cab5721374751c51b0 in mainline linux | |
|
7eb2eeb0 | 2025-05-23 02:32:31 | drm/amd: Add Suspend/Hibernate notification callback support From Mario Limonciello 4d45a5f1e2a04f5f84d5d5121dfbfff8d65c9c1c in linux-6.12.y/6.12.30 2965e6355dcdf157b5fafa25a2715f00064da8bf in mainline linux | |
|
c9b0513d | 2025-05-23 02:26:42 | add register_pm_notifier() for 6.12.30 drm | |
 |
ba45935f | 2025-05-22 10:50:10 | ix(4): use 4k clusters in LRO mode to avoid m_defrag calls ok bluhm | |
|
90a7f92b | 2025-05-22 09:15:48 | Bring up multiple transmit and receive queues. The IAVF spec only says we have at least 5 vectors, so limit to 4 queues. We're not setting a proper RSS key or lookup table yet, so traffic is not distributed across queues correctly. from Yuichiro NAITO | |
|
134c5a4e | 2025-05-22 08:32:50 | pass correct sizes to free() calls in ice_free_vsi_qmaps() | |
|
0814c25f | 2025-05-22 06:41:20 | Move packet classifier defines to if_ixlreg.h so iavf(4) can see them too. ok dlg@ | |
|
85baac77 | 2025-05-22 06:34:03 | The current way to adjust pf(4) limits in pf.conf(5) is inconvenient. For example when ruleset uses more than 512 anchors (the current default limit) one would typically add 'set limit anchor 1024' to adjust the limit so the 'pf.conf(5)' gets processed. Unfortunately it does not work because limit gets changed with DIOCXCOMMIT which is too late. The pf.conf(5) fails to load the anchors to transaction, because the old lower limit is still in place. To fix it we must set the limit as soon as we parse 'set limit ...' option. The issue has been reported and fix tested by rafal _dot_ ramocki _von_ eo.pl OK @bluhm | |
 |
40adde36 | 2025-05-22 03:12:33 | Fix trailing whitespace. | |
|
f4df864d | 2025-05-22 03:09:00 | Remove redundant NULL check from divert_packet() that it already in in_pcbunref(). | |
 |
0487f6a6 | 2025-05-22 03:04:01 | Add PERFORMANCE_LEVEL_SET logic and initialize all domains to the same level. On Snapdragon X Elite there are 3 performance domains spanning 4 CPUs each, where each can go from level 0 (719 Mhz) to 12 (3418 Mhz). By default domain 0 boots into level 10, domains 1 and 2 boot into level 12. Since the kernel only understands a single global performance value we keep them in sync by initializing them all to 12 for now. While there, also move PERFORMANCE_LEVEL_GET to a separate function and clean up a bit. ok patrick@ | |
 |
c902741c | 2025-05-21 18:41:41 | Define CPU_MIN_BUSY_CYCLES and CPU_MAX_BUSY_CYCLES outside of MULTIPROCESSOR since the ddb_mtx_enter is not under MULTIPROCESSOR. This is enough to unbreak GENERIC kernels. Reported by tb@ | |
 |
fdb740e5 | 2025-05-21 16:59:32 | Zap uvm_share(). No longer used by vmm(4). ok mpi@ | |
 |
d6f52495 | 2025-05-21 14:10:16 | kqueue: Improve EVFILT_USER compatibility Always set `data' and `udata' of the EVFILT_USER event when triggering it with NOTE_TRIGGER. Preserve `fflags' when the event is cleared with EV_CLEAR. These make the behaviour more similar to kqueue(2) / kevent(2) of FreeBSD, to avoid OS-specific quirks in user code. Tested by volker@ OK tedu@ mvs@ | |
|
c2568eef | 2025-05-21 12:51:14 | remove unused HTOLE16 macro that somehow came over here from ixl(4) | |
|
e03501a6 | 2025-05-21 12:45:27 | Don't process link state change events that arrive before we've fully set up the interface. from Yuichiro NAITO | |
 |
1f62937d | 2025-05-21 09:42:59 | In preparation of moving work from the reaper into exit1(), introduce a new (optional) pmap function that "purges" a pmap to optimize a subsqeuent uvm space teardown. Rewire the arm64 TLB flush optimization to use that new function. ok mpi@, claudio@ | |
 |
fa0a14c8 | 2025-05-21 09:33:48 | Get rid of unused `pr_hardlimit_warning', `pr_hardlimit_ratecap' and `pr_hardlimit_warning_last'. ok dlg tedu | |
 |
ee0e82c9 | 2025-05-21 09:06:58 | Move pmap_deactivate() and sched_exit() into MI exit1(). ok kettenis@ | |
|
8e23f399 | 2025-05-21 09:04:00 | Improve mtx_enter() for machines with a huge number of CPUs. - Always read the value just before issuing an atomic operation to avoid locking the cache line. - Use an exponential backoff for the busy loop when waiting for a contended lock. Prevent hangs on Ampere Altra with highly contended mutexes. ok kettenis@, dlg@ | |
|
6a2e23f1 | 2025-05-21 04:13:52 | Add acpitimer so we get a working delay function on some (virtual) machines that lack acpihpet. This fixes crashes when installing from emulated IDE CD drives, as reported most recently by Nicole Findlay. ok mlarkin@ | |
 |
94254965 | 2025-05-21 04:11:57 | remove a couple includes that weren't needed from previous commit The previous commit had an include added that wasn't needed. | |
|
1969dc37 | 2025-05-21 04:10:21 | Move the code to attach acpitimer outside #ifndef SMALL_KERNEL so we can make the clock/delay setup on ramdisk kernels more like GENERIC. ok mlarkin@ | |
|
579b2e07 | 2025-05-21 04:05:22 | preallocate hibernate work area during boot Attempt to preallocate the work area needed for hibernate during boot, instead of waiting to do this immediately before hibernate. Fixes some scenarios where a suitably large and correctly aligned region cannot be late-allocated. tested by many ok krw | |
 |
d117d18f | 2025-05-21 02:18:29 | only register the notify function if we actually have something to handle | |
|
267a06d5 | 2025-05-21 00:13:44 | establish the dt_deferred_wakeup() softintr as MPSAFE. without this, profiling on a busy system with btrace ends up with all the cpus spinning on the kernel lock all the time when the dt event buffer gets full. when the event buffer is full, dt tries to wake up btrace to read the buffer, but it defers this wakup to a softintr to avoid calling wakeup() at really high IPLs. because dt hides itself from stack traces, it looks ambiguous where this contention comes from and it's often assumed that it is the timeout subsystem that's causing the problems. we have improved timeouts, but apart from optimising timeout_barrier out of the picture the flame graphs looked basically the same as before. this is possible now that softintrs are implemented consistently across our archs (thanks visa@). tested by and ok bluhm@ | |
|
1363fb03 | 2025-05-20 18:41:06 | Unlock TCPCTL_REASS_LIMIT and TCPCTL_SACKHOLE_LIMIT cases of tcp_sysctl(). Use the pool lock to serialize pool_sethardlimit() with the rest pool layer. Also use `sysctl_lock' to serialize pool and sysctl variable modification. Since the whole tcp_sysctl() became mp-safe, move it out of sysctl locks. ok bluhm | |
|
b8a5eea1 | 2025-05-20 18:40:09 | Move move IPCTL_SOURCEROUTE case of ip_sysctl() out of netlock. It is atomically accessed integer. sysctl_securelevel_int() is mp-safe. ok bluhm | |
|
bde04a5b | 2025-05-20 13:51:27 | Back vmm(4) guest memory with UVM aobjs. For the past few years, vmm(4) would require vmd(8) to allocate virtual memory via mmap(2) in the vm process to use as guest physical memory. Then vmm(4) would create a new virtual address space itself to represent the guest and use uvm_share() to share map entries between them. This worked, but the userland dance of mmap/munmap to "find" space for guest memory and then having to teardown a full virtual address space not tied to a process caused some havoc in state management and potential race conditions. This commit simplifies how guest memory is represented and managed: 1. vmd(8) no longer calls mmap(2) to find memory for a vm. vmm(4) creates UVM aobjs to represent the memory ranges and maps them into the vmd process. 2. vmm(4) no longer faults pages into the vm's address space managed in the kernel and instead faults directly into the vmd userland process's address space. Nested page tables (EPT/RVI) are managed explicitly with pmap(9) functions. 3. vmd(8) shares guest memory between processes (e.g. virtio network & block devices) using the same ioctl as before, but vmm(4) now uses uvm_map(9) to map the aobj's into the new process. 4. The aobj mappings into the userland processes are now marked immutable. Tested with help from many: bluhm@, kirill@, hshoexer@, Jesper Wallin, xse[at]krkrkr.org. Thanks to Mischa Peters for providing access to an additional testing environment. Help from mpi@ on uvm aobj lifecycle management. ok mlarkin@ | |
|
14ec4249 | 2025-05-20 12:46:52 | remove unused pmap_remove_all() ok kettenis@ claudio@ | |
|
ac2f7dbf | 2025-05-20 10:02:32 | As done in ixl(4), set IFM_FDX when the link is active. from NetBSD if_iavf.c r1.13 | |
|
b877ea7a | 2025-05-20 09:43:31 | Add support for TSO to iavf(4), much like it's done in ixl(4). from Yuichiro NAITO, with some additions by me | |
|
75dab2ae | 2025-05-20 09:12:03 | add Rx checksum offload support to ice(4) based on code from FreeBSD test + ok bluhm@ | |
|
f2028685 | 2025-05-20 08:35:37 | Special vmentry/vmexit path for AMD SEV-ES guests in vmm(4). With SEV-ES the full vCPU state is automatically loaded from or saved to the encrypted VMSA. However, host state is not fully saved and restored. Therefore, we need a seperate vm entry/exit path for SEV-ES enabled guests. svm_seves_enter_guest() accomplishes this. from hshoexer@; OK mlarkin@ | |
|
e5b99936 | 2025-05-20 07:02:20 | Revert UVM_LK_* flag removal. | |
|
23f18c79 | 2025-05-20 05:51:43 | Call in_pcbselsrc() and in6_pcbselsrc() with const sockaddr parameter. Functions in_pcbselsrc() and in6_pcbselsrc() use the destination sockaddr to determine a suitable source address. As the destination is only used for lookup, it should never be modified. Use const to let the compiler verify that. On the way down the callstack, also convert a bunch of other lookup functions to const. OK kn@ | |
|
ce551b50 | 2025-05-20 05:50:18 | ip6_mroute: add missing counters ok bluhm | |
|
9a08a183 | 2025-05-20 02:06:54 | check if FnLock is available and enabled in the BIOS before fiddling with it | |
 |
6f3c3747 | 2025-05-20 01:22:39 | supposed to use init function in target | |
|
baa6eae0 | 2025-05-20 01:18:16 | initial cleanup run and use DPRINTF to be less noisy | |
|
dfc95de7 | 2025-05-20 01:10:42 | For AMD SEV-ES use GUEST_INTERRUPT_MASK in vmm(4). Due to state encryption vmm(4) cannot determine the state of the interrupt enable flag from the rflags register of the guest. Therefore the VMCB provides at offset 0x68 bit 1 the current state of this flag. However, use the SMV_GUEST_INTR_MASK bit only if this is a SEV-ES enabled guest. Otherwise, use the regular FLAGS register. from hshoexer@; OK mlarkin@ | |
|
2ec1c143 | 2025-05-20 00:37:17 | Run IPv6 fragment reassembly in parallel. frag6_input() is MP-safe, a global mutex is protecting it. Set PR_MPINPUT protocol flag to run it in parallel from ip_deliver() loop. Note that this only affects configurations where pf(4) has been turned off, as pf does its own fragment reassembly. OK mvs@ | |
|
c92d48cd | 2025-05-19 21:48:28 | Implement a ddb.suspend sysctl that may help debugging suspend/resume bugs. When this sysctl is set to 1, it will force "S0ix" suspend and skip suspend of inteldrm(4) and amdgpu(4) such that the display remains on during suspend. That should make anything printed on the console to be visible. Power consumption will probably suck though. ok mlarkin@, kn@ | |
|
fc9ae68f | 2025-05-19 19:15:19 | add a basic WMI driver, with support for asus laptops. now let some others fix the bugs! | |
|
56dcc2b2 | 2025-05-19 10:10:13 | remove unused bufcache_getdmacleanbuf() ok deraadt@ mpi@ beck@ | |
|
de414f7e | 2025-05-19 08:36:36 | Configure AMD SEV-ES in vmm(4). For SEV-ES we have to adjust the guest configuration: - Do not intercept XSETBV. We can not force access to XCR0 as it is part of the encrypted state. - We do not have direct access to EFER and CR[04], thus intercept EFER and CR[04] "post write". The provided exit handler will keep track of the guest state. - Regarding MSRs: - Allow access to GHCB MSR, which will be used for guest-host communication. - Allow reading XSS state, which will be needed for CPUID Extended State Enumeration. - Allow full read/write for EFER. SVME bit cannot not be modified with SEV-ES, it will always be 1. - SEV-ES requires LBR virtualization, thus enable it. When everything is set up, the initial state is copied to the VMSA, which is not yet encrypted. This will have to be done by vmd(8) and psp(4). from hshoexer@; OK mlarkin@ | |
 |
0e0b6be9 | 2025-05-19 07:34:21 | goto bad if we do not have a route. This allows us to unindent a whole block. Suggested by & OK kn | |
|
206922f6 | 2025-05-19 06:50:00 | nd6log is gone, mop up nd6_debug sysctl. input & OK kn, OK bluhm | |
|
64472281 | 2025-05-19 06:48:16 | We are no longer doing icmp6 redirect interop testing. nd6log has outlived its usefulness. input & OK kn, OK bluhm | |
|
bb279c7c | 2025-05-19 06:46:58 | icmp6_reflect has worked for some time, we are not debugging it. input & OK kn, OK bluhm | |
|
45132001 | 2025-05-19 06:46:24 | Embedding scope just works. input & OK kn, OK bluhm | |
|
5cc13a7e | 2025-05-19 06:45:49 | We are no longer doing icmp6 interop testing. nd6log has outlived its usefulness. input & OK kn, OK bluhm | |
|
9fd9dabb | 2025-05-19 06:45:14 | We might run out of memory, nd6log telling us is not helpful. input & OK kn, OK bluhm | |
|
fef37700 | 2025-05-19 06:44:38 | Not much we can do if we can't join a multicast group. Logging with nd6log is not helping anything. input & OK kn, OK bluhm | |
|
787133f8 | 2025-05-19 06:44:10 | Trust that we create correct route entries. input & OK kn, OK bluhm | |
|
4eeeb7fa | 2025-05-19 06:43:28 | We are no longer doing ND option interop testing. nd6log has outlived its usefulness. input & OK kn, OK bluhm | |
|
4bab435d | 2025-05-19 06:42:53 | Duplicate address detection has been working for a few decades. Get rid of nd6log, it's no longer useful. input & OK kn, OK bluhm | |
|
79dff0e6 | 2025-05-19 06:42:13 | There is not much we can do if we can't find our source address. Logging with nd6log is not helping anything. input & OK kn, OK bluhm | |
|
c6663b6f | 2025-05-19 06:41:31 | We are no longer doing neighbour advertising interop testing. IPv6 is reasonably mature that we can just drop invalid packets, nd6log has outlived its usefulness. Pull down the inet_ntop(3) buffer into its own scope where it's needed. input & OK kn, OK bluhm | |
|
f1e4bcef | 2025-05-19 06:40:59 | There is not much we can do if we can't find our source address. Logging with nd6log is not helping anything. input & OK kn, OK bluhm | |
|
f6010121 | 2025-05-19 06:40:18 | We are no longer doing neighbour solicitation interop testing. IPv6 is reasonably mature that we can just drop invalid packets, nd6log has outlived its usefulness. input & OK kn, OK bluhm | |
|
57af335d | 2025-05-19 06:39:19 | We are no longer doing neighbour discovery interop testing. IPv6 is reasonably mature that we can just drop invalid packets, nd6log has outlived its usefulness. input & OK kn, OK bluhm | |
|
1dc06558 | 2025-05-19 06:38:33 | We are no longer debugging behaviour of tentative or duplicate addresses. It's been working just fine for decades, nd6log has outlived its usefulness. input & OK kn, OK bluhm | |
|
ef814849 | 2025-05-19 06:36:48 | Stop logging when rt_ifa_del fails, nobody else does. input & OK kn, OK bluhm | |
|
aabd3c3e | 2025-05-19 06:36:06 | Remove nd6log from in6_update_ifa() * Sending a vltime of 0 is no longer a typical configuration mistake of a tool's bug. Also we are no longer handling router advertisements in the kernel, this belongs in slaacd(8). * To be able to remove a char[] buffer, stop logging that rt_ifa_del failed, most other callers just quietly putter along if it fails. input & OK kn, OK bluhm | |
|
56a449ad | 2025-05-19 06:34:54 | We are no longer debugging link-local address generation on a regular basis. nd6log has outlived its usefulness. input & OK kn, OK bluhm | |
|
0c8363b4 | 2025-05-19 04:54:04 | Use per CPU counter for IPv6 multicast stats. ok bluhm@ | |
|
f4fc7468 | 2025-05-19 02:27:57 | Disable TCP softlro for small kernels. Saves 6.7K object size. OK kn@ jan@ | |
|
f2f0070e | 2025-05-19 02:13:15 | remove unused NIRQ defines; ok mlarkin@ | |
|
88f9fc2b | 2025-05-18 23:27:29 | ip6_mroute: remove unused parameter from mf6c_find and mrt6_mcast_add ok mvs | |
|
4d3d2000 | 2025-05-18 23:26:30 | drm/amdgpu/hdp7: use memcfg register to post the write for HDP flush From Alex Deucher 92d0a28afba2166801ed563a72aed37cb8ea27f7 in linux-6.12.y/6.12.29 5a11a2767731139bf87e667331aa2209e33a1d19 in mainline linux | |
|
5c984ccf | 2025-05-18 23:24:45 | drm/amdgpu/hdp6: use memcfg register to post the write for HDP flush From Alex Deucher df044182621ae2632e89d054835ddf23f64e24a5 in linux-6.12.y/6.12.29 ca28e80abe4219c8f1a2961ae05102d70af6dc87 in mainline linux | |
|
9145c2f4 | 2025-05-18 23:23:11 | drm/amdgpu/hdp5: use memcfg register to post the write for HDP flush From Alex Deucher 0a776c305435e24ad0f52dc142b7822bd862de95 in linux-6.12.y/6.12.29 0e33e0f339b91eecd9558311449a3d1e728722d4 in mainline linux | |
|
a70f42f5 | 2025-05-18 23:21:39 | drm/amdgpu/hdp5.2: use memcfg register to post the write for HDP flush From Alex Deucher b6f0f3e6c7a901c4bc7b5198276f9c96e0e8200a in linux-6.12.y/6.12.29 dbc988c689333faeeed44d5561f372ff20395304 in mainline linux | |
|
9161f22d | 2025-05-18 23:19:39 | drm/amdgpu/hdp4: use memcfg register to post the write for HDP flush From Alex Deucher 4872de413e20a9e151e8cca0f9f4603c056ee1be in linux-6.12.y/6.12.29 f690e3974755a650259a45d71456decc9c96a282 in mainline linux |