hooks/pre-receive

   1 #!/bin/sh
   2 #
   3 # Keep track of the last time we modified the object store
   4 #
   5 # Beware, we MAY be running in a chroot!
   6
   7 set -e
   8
   9 # Make sure the current directory is where we expect to be
  10 [ "${GIT_DIR+set}" != "set" ] || cd "${GIT_DIR:-.}" || exit 1
  11 case "${PWD%/*}" in */worktrees)
  12         # Gah!
  13         if
  14                 # But it COULD just be a coincidence...
  15                 [ -s commondir ] && [ -s HEAD ] &&
  16                 _cmndir= && read -r _cmndir <commondir 2>/dev/null &&
  17                 [ -n "$_cmndir" ] && [ -d "$_cmndir" ]
  18         then
  19                 # ...it is not, fix it!
  20                 cd "$_cmndir" || exit 1
  21         fi
  22 esac
  23 [ "${GIT_DIR+set}" != "set" ] || GIT_DIR="."
  24
  25 # Get out of the mob
  26 case "$PWD" in *?/mob)
  27         cd ..
  28         GIROCCO_PERSONAL_MOB=1
  29 esac
  30
  31 umask 002
  32
  33 if [ -x @perlbin@ ]; then
  34         # We are NOT inside the chroot
  35         basedir=@basedir@
  36         list_packs() { command "$basedir/bin/list_packs" "$@"; }
  37         strftime() { command "$basedir/bin/strftime" "$@"; }
  38 fi
  39
  40 git config gitweb.lastreceive "$(date '+%a, %d %b %Y %T %z')"
  41
  42 # Read the incoming refs and freshen old loose objects
  43 # If we waited until post-receive a gc could have already nuked them
  44 # We freshen the new ref in case it's being resurrected to protect it from gc
  45 # We probably do not need to do it for new refs as Git tries to do that,
  46 # but since we're already doing it for old refs (which Git does not do),
  47 # it's almost no extra work for new refs, just in case.  We also attempt to
  48 # make all packs user and group writable so they can be touched/renamed later.
  49
  50 # Starting with Git v2.11.0 receive-pack packs/objects end up in a quarantine
  51 # object directory that is just discarded immediately if pre-receive declines
  52 # to accept the push.  This is a good thing.  However, it means that the
  53 # incoming objects are NOT located in objects/... as GIT_OBJECT_DIRECTORY and
  54 # GIT_QUARANTINE_PATH are both set to the quarantine objects directory and the
  55 # original objects directory is appended to GIT_ALTERNATE_OBJECT_DIRECTORIES
  56 # (but it will just be the absolute path to objects).  The simple bottom line
  57 # is that we should also try everything in the GIT_QUARANTINE_PATH directory if
  58 # it's set.
  59 [ -z "$GIT_QUARANTINE_PATH" ] || [ -d "$GIT_QUARANTINE_PATH" ] || unset GIT_QUARANTINE_PATH
  60
  61 # We also record changes to a ref log.  We do it here rather than in
  62 # post-receive so that we can guarantee all potential changes are recorded in
  63 # the log before they take place.  It's possible that the update hook will
  64 # ultimately deny one or more updates but waiting until post-receive could
  65 # result in updates being left out of the log.
  66
  67 hexdig='[0-9a-f]'
  68 octet="$hexdig$hexdig"
  69 octet4="$octet$octet$octet$octet"
  70 octet20="$octet4$octet4$octet4$octet4$octet4"
  71 _make_packs_ugw() {
  72         find -L "$1" -maxdepth 1 -type f ! -perm -ug+w \
  73                 -name "pack-$octet20*.*" -exec chmod ug+w '{}' + || :
  74 } 2>/dev/null
  75 _make_packs_ugw objects/pack
  76 [ -z "$GIT_QUARANTINE_PATH" ] || _make_packs_ugw "$GIT_QUARANTINE_PATH/pack"
  77
  78 # Trigger a mini-gc if there are at least 20 packs present.
  79 # Our current pack that contains this push's data will have a .keep while
  80 # this hook is running so it will be excluded from the count, but the count
  81 # is only an approximate guide anyway and being off by one or two will not
  82 # cause any performance problems.  We could drop the check to 19 instead to
  83 # compensate but there could be other simultaneous pushes and it's highly
  84 # unlikely that it would end up making any difference anyway.  We also
  85 # deliberately ignore anything in the "$GIT_QUARANTINE_PATH/pack" area because
  86 # at this point it will have a .keep (if it exists) and because of the way
  87 # it's created there can never be more than one pack in there anyway.
  88 # The mini-gc code contains the logic to sort out small packs vs. non-small
  89 # packs and which should be combined in what order so we do not need to
  90 # do any more complicated testing here.
  91 if ! [ -e .needsgc ]; then
  92         packs=
  93         { packs="$(list_packs --quiet --count --exclude-no-idx --exclude-keep objects/pack || :)" || :; } 2>/dev/null
  94         if [ -n "$packs" ] && [ "$packs" -ge 20 ]; then
  95                 >.needsgc
  96         fi
  97 fi
  98
  99 # Make sure we have a reflogs directory and abort the update if we cannot
 100 # create one.  Normally project creation will make sure this directory exists.
 101 [ -d reflogs ] || mkdir -p reflogs >/dev/null 2>&1 || :
 102 [ -d reflogs ]
 103
 104 # Multiple push operations could be occurring simultaneously so we need to
 105 # guarantee they do not step on one another and we do this by generating a
 106 # unique log file name.  We use a microseconds timestamp and the current process
 107 # id and we guarantee that this process is kept alive for the entire microsecond
 108 # of the timestamp thereby guaranteeing that we are the only possible process
 109 # that could use that pid during that particular microsecond (ignoring leap seconds).
 110 # To do this we need to sleep until the microsecond turns over, grab the timestamp
 111 # and then sleep until the microsecond turns over again.  This will introduce a
 112 # guaranteed 2 microsecond delay into every push.  This should not generally be
 113 # noticeable.  The strftime utility does this for us when using %N with current time.
 114 # We always use UTC for the timestamp so that chroot and non-chroot match up.
 115 # Log entries are the lines sent to the pre-receive hook with hhmmss prepended.
 116 lognamets="$(TZ=UTC strftime '%Y%m%d_%H%M%S%N')"
 117 lognamets="${lognamets%???}"
 118 loghhmmss="${lognamets##*_}"
 119 loghhmmss="${loghhmmss%??????}"
 120
 121 # We write to a temp ref log and then move it into place so that the reflogs
 122 # collector can assume that log files with their final name are immutable
 123 logname="reflogs/$lognamets.$$"
 124 lognametmp="reflogs/tmp_$lognamets.$$"
 125
 126 while read -r old new ref; do
 127         echo "$loghhmmss $old $new $ref" >&3
 128         args=
 129         if [ "$old" != "0000000000000000000000000000000000000000" ]; then
 130                 # freshen mod time on recently unref'd loose objects
 131                 fn="${old#??}"
 132                 shard="${old%$fn}"
 133                 args="$args 'objects/$shard/$fn'"
 134                 [ -z "$GIT_QUARANTINE_DIRECTORY" ] || args="$args '$GIT_QUARANTINE_DIRECTORY/$shard/$fn'"
 135         fi
 136         if [ "$new" != "0000000000000000000000000000000000000000" ]; then
 137                 # prevent imminent pruning of a ref being resurrected
 138                 fn="${new#??}"
 139                 shard="${new%$fn}"
 140                 args="$args 'objects/$shard/$fn'"
 141                 [ -z "$GIT_QUARANTINE_DIRECTORY" ] || args="$args '$GIT_QUARANTINE_DIRECTORY/$shard/$fn'"
 142         fi
 143         eval "chmod ug+w $args" 2>/dev/null || :
 144         eval "touch -c $args" 2>/dev/null || :
 145 done 3>"$lognametmp"
 146 mv "$lognametmp" "$logname"
 147
 148 # While unlikely, it is conceivable that several ref updates have occurred that
 149 # did not actually push any packs.  In that case we could build up a large
 150 # number of log files so request a mini gc if there are 50 or more of them now.
 151 if ! [ -e .needsgc ]; then
 152         logfiles=
 153         { logfiles="$(($(find -L reflogs -maxdepth 1 -type f -print | wc -l || :)+0))" || :; } 2>/dev/null
 154         if [ -n "$logfiles" ] && [ "$logfiles" -ge 50 ]; then
 155                 >.needsgc
 156         fi
 157 fi