BSD Commit Log Search

   led(4): minor tweaks

   envctrl(4): touch up

   XXX: mandoc seems to be buggy and does not evaluate .Bl -column
   strings, so the columns are too wide in mandoc plain text ouput.

   gcc/arm: For -mtp=soft, ensure stack alignment even in leaves.

   The option -mtp=soft, which is the default on earmv5, makes queries
   to the thread pointer, for access to static (`initial-exec')
   thread-local storage, go through the C runtime subroutine
   __aeabi_read_tp.  (For earmv>=6, we use the cp15 register via a
   single instruction.)

   Thus procedures which gcc thinks of as leaf procedures that use
   __aeabi_read_tp are not really leaf procedures -- and even though
   __aeabi_read_tp itself doesn't use the stack pointer at all,
   resolving the symbol may take a detour through the dynamic linker,
   which does rely on an aligned stack pointer.

   Without this change, code like

   __thread int x = 0;
   int *
   get_x(void)

    [92 lines not shown]

   The 2nd argument of fdt_memory_remove_range is a size. Fix the call sites
   that get this wrong.

   Prompted by the email to port-riscv from Alex Jokela

   doc: Two weeks of changes.

   Update the envctrl manual page to reflect changes:
   - remove history/authors because the driver has changed
   - add E250 information

   Add build glue for the led.4 (generic LED framework) manual page.

   Add (belatedly) a manual page for the generic LED framework.

   fil.c: fix typo in a comment

   gdb: Disable emoji in output by default.

   NetBSD users may be more likely than others to be using a physical
   DEC VT420 as their console, or at the very least the standard x86
   BIOS text console while debugging the system. Both of these would
   rather display a garbled mess than an emoji.

   Besides, this is the system debugger, so we want to keep things
   usable in a minimal environment.

   Users can still re-enable emoji output with "set style emoji on"
   if they are using kalacrittyology7 with an utf-8 locale. Or uxterm
   with symbola-ttf installed (monochrome only).

   PR bin/60497 gdb prints garbled gibberish in an attempt to show
   cutesy emoji

   Remove unnecessary #includes

   Note cobalt, shark and sparc64 mcclock attachments.  Mention MC146816.
   New line after full stop whilst here.

   turn off sync signals on both CRTCs when blanking

   Un-hide basically everything.

   hdaudio(4): Apply access checks to ioctl routines.

   This way the permissions on /dev/hdaudioN actually do something.

   Sort includes while here.

   PR kern/60492: hdaudio(4): setconfig is allowed unprivileged

   arp: Validate op on input, before creating any state.

   PR kern/60491: arp creates state for invalid messages

   nd6: Redo previous avoidance of buffer overread.

   Previous change invalidated two checks of the length:

       285      if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
       286              nd6log(LOG_INFO, "lladdrlen mismatch for %s "
       287                  "(if %d, NS packet %d)\n",
       288                  IN6_PRINT(ip6buf, &taddr6),
       289                  ifp->if_addrlen, lladdrlen - 2);
       290              goto bad;
       291      }

   https://nxr.netbsd.org/xref/src/sys/netinet6/nd6_nbr.c?r=1.186#285

       685              if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
       686                      nd6log(LOG_INFO, "lladdrlen mismatch for %s "
       687                          "(if %d, NA packet %d)\n",
       688                          IN6_PRINT(ip6buf, &taddr6),
       689                          ifp->if_addrlen, lladdrlen - 2);

    [18 lines not shown]

   nd6: Avoid buffer overread of link-layer address options.

   PR kern/60488: nd6: two-byte buffer overread in source/target
   linkaddr options

   netinet6: Plug leak and avoid arith overflow in fragment accounting.

   kern/60487: netinet6: fragment accounting leak

   tcp syncache: Don't leak kernel stack data through timestamps.

   kern/60486: tcp syncache leaks 4-byte kernel stack content

   sun4v vdsk: catch up with recent updates to cd(4)

   Add support for missing SCSI_MODE_SELECT_10 and  SCSI_MODE_SENSE_10.
   Verified to work in a sun4v ldom with acccess to a virtual cd.

   pf(4): Avoid UAF of pf mbuf tag in fragment reassembly.

   PR kern/60485: pf(4): uaf in fragment reassembly

   ipfilter: Don't choke on an ICMP6 error quoting an ICMP6 error.

   PR kern/60484: ipfilter: null pointer deref in icmp6 parsing

   ipfilter: Test icmp6 packet of death.

   PR kern/60484: ipfilter: null pointer deref in icmp6 parsing

   aic79xx.c: fix spelling and punctuation consistency

   (Not sure why either of those terms warrant quotation marks, but the
   existing style was retained.)

   MAXPARTITIONS is required to build libc, so make it visible outside the
   kernel.

   Welcome to 11.0 RC7

   PR kern/60477

   Use greater than zero instead of strictly one

   PR kern/60477

   Use the right assertions to ensure that all IOs are suspended.
   zero indicates that IOs resume.

   ld.elf_so: Prune some dead TLS variant branches.

   __lwp_settcb / __lwp_gettcb_fast are used only with TLS variant I,
   never with TLS variant II.

   So say `#ifdef __HAVE_TLS_VARIANT_II' instead of `#ifndef
   __HAVE_TLS_VARIANT_I' to make it clearer, and prune all the
   __HAVE___LWP_SETTCB / __HAVE___LWP_GETTCB_FAST branches under it
   because they're unreachable.

   No functional change intended: all the deleted code should be dead.

   PR bin/60469: bin/60469: assertion "ALIGNED_P(q, obj->tlsalign)"
   failed: file "/usr/src/libexec/ld.elf_so/tls.c", line 333