cvs commit: src/sys/kern subr_witness.c src/sys/netinet igmp.c in.c in_var.h ip_input.c ip_output.c src/sys/netinet6 in6_var.h mld6.c

[Ed Maste]

emaste      2006-01-18 23:36:49 UTC

  FreeBSD src repository

  Modified files:        (Branch: RELENG_5)
    sys/kern             subr_witness.c 
    sys/netinet          igmp.c in.c in_var.h ip_input.c 
                         ip_output.c 
    sys/netinet6         in6_var.h mld6.c 
  Log:
  Merge IP multicast address list locking from HEAD to RELENG_5, with a
  few changes to preserve ABI compatibility.
  
  subr_witness.c:1.196:
    Add in_multi_mtx, igmp_mtx, and if_addr_mtx lock order to hard-coded
    lock order in WITNESS, in that order.
  
  in_var.h:1.54 in6_var.h:1.23 mld6.c:1.22:
    Modify network protocol consumers of the ifnet multicast address lists
    to lock if_addr_mtx.
  
  igmp.c:1.49 in.c:1.86: in_var.h:1.55 ip_input.c:1.303 ip_output.c:1.243:
    Introduce in_multi_mtx, which will protect IPv4-layer multicast address
    lists, as well as accessor macros.  For now, this is a recursive mutex
    due code sequences where IPv4 multicast calls into IGMP calls into
    ip_output(), which then tests for a multicast forwarding case.
  
    For support macros in in_var.h to check multicast address lists, assert
    that in_multi_mtx is held.
  
    Acquire in_multi_mtx around iteration over the IPv4 multicast address
    lists, such as in ip_input() and ip_output().
  
    Acquire in_multi_mtx when manipulating the IPv4 layer multicast
    addresses, as well as over the manipulation of ifnet multicast address
    lists in order to keep the two layers in sync.
  
    Lock down accesses to IPv4 multicast addresses in IGMP, or assert the
    lock when performing IGMP join/leave events.
  
    Eliminate spl's associated with IPv4 multicast addresses, portions of
    IGMP that weren't previously expunged by IGMP locking.
  
  in.c:1.89 in.c:1.90:
    Acquire Giant conditionally in in_addmulti() and in_delmulti() based on
    whether the interface being accessed is IFF_NEEDSGIANT or not.  This
    avoids lock order reversals when calling into the interface ioctl
    handler, which could potentially lead to deadlock.
  
    The long term solution is to eliminate non-MPSAFE network drivers.
  
    Discussed with: jhb
  
    Unlock Giant symmetrically with respect to lock acquire order as that's
    generally nicer.
  
    Spotted by:     johan
  
  ip_output.c:1.244:
    Add helper function ip_findmoptions(), which accepts an inpcb, and attempts
    to atomically return either an existing set of IP multicast options for the
    PCB, or a newlly allocated set with default values.  The inpcb is returned
    locked.  This function may sleep.
  
    Call ip_moptions() to acquire a reference to a PCB's socket options, and
    perform the update of the options while holding the PCB lock.  Release the
    lock before returning.
  
    Remove garbage collection of multicast options when values return to the
    default, as this complicates locking substantially.  Most applications
    allocate a socket either to be multicast, or not, and don't tend to keep
    around sockets that have previously been used for multicast, then used for
    unicast.
  
    This closes a number of race conditions involving multiple threads or
    processes modifying the IP multicast state of a socket simultaenously.
  
  Approved by:    rwatson (mentor)
  
  Revision    Changes    Path
  1.178.2.9   +7 -0      src/sys/kern/subr_witness.c
  1.46.2.2    +12 -8     src/sys/netinet/igmp.c
  1.77.2.7    +27 -10    src/sys/netinet/in.c
  1.51.2.2    +16 -0     src/sys/netinet/in_var.h
  1.283.2.15  +2 -0      src/sys/netinet/ip_input.c
  1.225.2.13  +61 -33    src/sys/netinet/ip_output.c
  1.19.2.3    +2 -0      src/sys/netinet6/in6_var.h
  1.17.2.3    +2 -0      src/sys/netinet6/mld6.c