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 [ "${GIT_DIR+set}" != "set" ] || { [ -n "$GIT_DIR" ] && [ -d "$GIT_DIR" ]; } || unset GIT_DIR
  11 [ -n "$GIT_DIR" ] || GIT_DIR="$(git rev-parse --git-dir)"
  12 [ -n "$GIT_DIR" ] && cd -P "${GIT_DIR:-.}" || exit 1
  13 case "${PWD%/*}" in */worktrees)
  14         # Gah!
  15         if
  16                 # But it COULD just be a coincidence...
  17                 [ -s commondir ] && [ -s HEAD ] &&
  18                 _cmndir= && read -r _cmndir <commondir 2>/dev/null &&
  19                 [ -n "$_cmndir" ] && [ -d "$_cmndir" ]
  20         then
  21                 # ...it is not, fix it!
  22                 cd -P "$_cmndir" || exit 1
  23         fi
  24 esac
  25 GIT_DIR="." GIT_PREFIX= && export GIT_DIR
  26
  27 # Get out of the mob
  28 case "$PWD" in *?/mob)
  29         cd ..
  30         GIROCCO_PERSONAL_MOB=1
  31 esac
  32
  33 umask 002
  34
  35 if [ -x @perlbin@ ]; then
  36         # We are NOT inside the chroot
  37         basedir=@basedir@
  38         list_packs() { command "$basedir/bin/list_packs" "$@"; }
  39         strftime() { command "$basedir/bin/strftime" "$@"; }
  40 fi
  41
  42 git config gitweb.lastreceive "$(date '+%a, %d %b %Y %T %z')"
  43
  44 # Read the incoming refs and freshen old loose objects
  45 # If we waited until post-receive a gc could have already nuked them
  46 # We freshen the new ref in case it's being resurrected to protect it from gc
  47 # We probably do not need to do it for new refs as Git tries to do that,
  48 # but since we're already doing it for old refs (which Git does not do),
  49 # it's almost no extra work for new refs, just in case.  We also attempt to
  50 # make all packs user and group writable so they can be touched/renamed later.
  51
  52 # Starting with Git v2.11.0 receive-pack packs/objects end up in a quarantine
  53 # object directory that is just discarded immediately if pre-receive declines
  54 # to accept the push.  This is a good thing.  However, it means that the
  55 # incoming objects are NOT located in objects/... as GIT_OBJECT_DIRECTORY and
  56 # GIT_QUARANTINE_PATH are both set to the quarantine objects directory and the
  57 # original objects directory is appended to GIT_ALTERNATE_OBJECT_DIRECTORIES
  58 # (but it will just be the absolute path to objects).  The simple bottom line
  59 # is that we should also try everything in the GIT_QUARANTINE_PATH directory if
  60 # it's set.
  61 [ -z "$GIT_QUARANTINE_PATH" ] || [ -d "$GIT_QUARANTINE_PATH" ] || unset GIT_QUARANTINE_PATH
  62
  63 # We also record changes to a ref log.  We do it here rather than in
  64 # post-receive so that we can guarantee all potential changes are recorded in
  65 # the log before they take place.  It's possible that the update hook will
  66 # ultimately deny one or more updates but waiting until post-receive could
  67 # result in updates being left out of the log.
  68
  69 hexdig='[0-9a-f]'
  70 octet="$hexdig$hexdig"
  71 octet4="$octet$octet$octet$octet"
  72 octet20="$octet4$octet4$octet4$octet4$octet4"
  73 _make_packs_ugrw() {
  74         find -L "$1" -maxdepth 1 -type f ! -perm -ug+rw,o+r \
  75                 -name "pack-$octet20*.*" -exec chmod ug+rw,o+r '{}' + || :
  76 } 2>/dev/null
  77 _make_packs_ugrw objects/pack
  78 if [ -n "$GIT_QUARANTINE_PATH" ]; then
  79         # gc.sh will be unable remove stale incoming-* dirs without this
  80         chmod ug+rwx,o+rx "$GIT_QUARANTINE_PATH" "$GIT_QUARANTINE_PATH/pack" >/dev/null 2>&1 || :
  81         chmod g+s "$GIT_QUARANTINE_PATH" "$GIT_QUARANTINE_PATH/pack" >/dev/null 2>&1 || :
  82         _make_packs_ugrw "$GIT_QUARANTINE_PATH/pack"
  83 fi
  84
  85 # Trigger a mini-gc if there are at least 20 packs present.
  86 # Our current pack that contains this push's data will have a .keep while
  87 # this hook is running so it will be excluded from the count, but the count
  88 # is only an approximate guide anyway and being off by one or two will not
  89 # cause any performance problems.  We could drop the check to 19 instead to
  90 # compensate but there could be other simultaneous pushes and it's highly
  91 # unlikely that it would end up making any difference anyway.  We also
  92 # deliberately ignore anything in the "$GIT_QUARANTINE_PATH/pack" area because
  93 # at this point it will have a .keep (if it exists) and because of the way
  94 # it's created there can never be more than one pack in there anyway.
  95 # The mini-gc code contains the logic to sort out small packs vs. non-small
  96 # packs and which should be combined in what order so we do not need to
  97 # do any more complicated testing here.
  98 if ! [ -e .needsgc ]; then
  99         packs=
 100         { packs="$(list_packs --quiet --count --exclude-no-idx --exclude-keep objects/pack || :)" || :; } 2>/dev/null
 101         if [ -n "$packs" ] && [ "$packs" -ge 20 ]; then
 102                 >.needsgc
 103         fi
 104 fi
 105
 106 # Make sure we have a reflogs directory and abort the update if we cannot
 107 # create one.  Normally project creation will make sure this directory exists.
 108 [ -d reflogs ] || mkdir -p reflogs >/dev/null 2>&1 || :
 109 [ -d reflogs ]
 110
 111 # Multiple push operations could be occurring simultaneously so we need to
 112 # guarantee they do not step on one another and we do this by generating a
 113 # unique log file name.  We use a microseconds timestamp and the current process
 114 # id and we guarantee that this process is kept alive for the entire microsecond
 115 # of the timestamp thereby guaranteeing that we are the only possible process
 116 # that could use that pid during that particular microsecond (ignoring leap seconds).
 117 # To do this we need to sleep until the microsecond turns over, grab the timestamp
 118 # and then sleep until the microsecond turns over again.  This will introduce a
 119 # guaranteed 2 microsecond delay into every push.  This should not generally be
 120 # noticeable.  The strftime utility does this for us when using %N with current time.
 121 # We always use UTC for the timestamp so that chroot and non-chroot match up.
 122 # Log entries are the lines sent to the pre-receive hook with hhmmss prepended.
 123 lognamets="$(TZ=UTC strftime '%Y%m%d_%H%M%S%N')"
 124 lognamets="${lognamets%???}"
 125 loghhmmss="${lognamets##*_}"
 126 loghhmmss="${loghhmmss%??????}"
 127
 128 # We write to a temp ref log and then move it into place so that the reflogs
 129 # collector can assume that log files with their final name are immutable
 130 logname="reflogs/$lognamets.$$"
 131 lognametmp="reflogs/tmp_$lognamets.$$"
 132
 133 while read -r old new ref; do
 134         echo "$loghhmmss $old $new $ref" >&3
 135         args=
 136         if [ "$old" != "0000000000000000000000000000000000000000" ]; then
 137                 # freshen mod time on recently unref'd loose objects
 138                 fn="${old#??}"
 139                 shard="${old%$fn}"
 140                 args="$args 'objects/$shard/$fn'"
 141                 [ -z "$GIT_QUARANTINE_DIRECTORY" ] || args="$args '$GIT_QUARANTINE_DIRECTORY/$shard/$fn'"
 142         fi
 143         if [ "$new" != "0000000000000000000000000000000000000000" ]; then
 144                 # prevent imminent pruning of a ref being resurrected
 145                 fn="${new#??}"
 146                 shard="${new%$fn}"
 147                 args="$args 'objects/$shard/$fn'"
 148                 [ -z "$GIT_QUARANTINE_DIRECTORY" ] || args="$args '$GIT_QUARANTINE_DIRECTORY/$shard/$fn'"
 149         fi
 150         eval "chmod ug+w $args" 2>/dev/null || :
 151         eval "touch -c $args" 2>/dev/null || :
 152 done 3>"$lognametmp"
 153 mv "$lognametmp" "$logname"
 154
 155 # While unlikely, it is conceivable that several ref updates have occurred that
 156 # did not actually push any packs.  In that case we could build up a large
 157 # number of log files so request a mini gc if there are 50 or more of them now.
 158 if ! [ -e .needsgc ]; then
 159         logfiles=
 160         { logfiles="$(($(find -L reflogs -maxdepth 1 -type f -print | wc -l || :)+0))" || :; } 2>/dev/null
 161         if [ -n "$logfiles" ] && [ "$logfiles" -ge 50 ]; then
 162                 >.needsgc
 163         fi
 164 fi