X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FGit.pm;h=181026c7bc1b726ff562e1b488f3256a0bf98c87;hb=b80dccff433cc529040e3c11604d76f669570511;hp=a756684a70355b6435b26b753f1dc6593e5481fa;hpb=2d4403cf9972f8ae78aa52fe6ce7a01d9b6757c1;p=public-inbox.git diff --git a/lib/PublicInbox/Git.pm b/lib/PublicInbox/Git.pm index a756684a..181026c7 100644 --- a/lib/PublicInbox/Git.pm +++ b/lib/PublicInbox/Git.pm @@ -1,4 +1,4 @@ -# Copyright (C) 2014-2018 all contributors +# Copyright (C) 2014-2020 all contributors # License: GPLv2 or later # # Used to read files from a git repository without excessive forking. @@ -8,12 +8,24 @@ # There are also API changes to simplify our usage and data set. package PublicInbox::Git; use strict; -use warnings; -use POSIX qw(dup2); -require IO::Handle; -use PublicInbox::Spawn qw(spawn popen_rd); -use base qw(Exporter); +use v5.10.1; +use parent qw(Exporter); +use POSIX (); +use IO::Handle; # ->autoflush +use Errno qw(EINTR); +use File::Glob qw(bsd_glob GLOB_NOSORT); +use Time::HiRes qw(stat); +use PublicInbox::Spawn qw(popen_rd); +use PublicInbox::Tmpfile; +use Carp qw(croak); our @EXPORT_OK = qw(git_unquote git_quote); +our $PIPE_BUFSIZ = 65536; # Linux default +our $in_cleanup; + +use constant MAX_INFLIGHT => + (($^O eq 'linux' ? 4096 : POSIX::_POSIX_PIPE_BUF()) * 3) + / + 65; # SHA-256 hex size + "\n" in preparation for git using non-SHA1 my %GIT_ESC = ( a => "\a", @@ -48,10 +60,8 @@ sub git_quote ($) { sub new { my ($class, $git_dir) = @_; - my @st; - $st[7] = $st[10] = 0; # may contain {-tmp} field for File::Temp::Dir - bless { git_dir => $git_dir, st => \@st, -git_path => {} }, $class + bless { git_dir => $git_dir, alt_st => '', -git_path => {} }, $class } sub git_path ($$) { @@ -59,6 +69,11 @@ sub git_path ($$) { $self->{-git_path}->{$path} ||= do { local $/ = "\n"; chomp(my $str = $self->qx(qw(rev-parse --git-path), $path)); + + # git prior to 2.5.0 did not understand --git-path + if ($str eq "--git-path\n$path") { + $str = "$self->{git_dir}/$path"; + } $str; }; } @@ -67,10 +82,11 @@ sub alternates_changed { my ($self) = @_; my $alt = git_path($self, 'objects/info/alternates'); my @st = stat($alt) or return 0; - my $old_st = $self->{st}; - # 10 - ctime, 7 - size - return 0 if ($st[10] == $old_st->[10] && $st[7] == $old_st->[7]); - $self->{st} = \@st; + + # can't rely on 'q' on some 32-bit builds, but `d' works + my $st = pack('dd', $st[10], $st[7]); # 10: ctime, 7: size + return 0 if $self->{alt_st} eq $st; + $self->{alt_st} = $st; # always a true value } sub last_check_err { @@ -92,140 +108,219 @@ sub _bidi_pipe { } return; } - my ($in_r, $in_w, $out_r, $out_w); - - pipe($in_r, $in_w) or fail($self, "pipe failed: $!"); + my ($out_r, $out_w); pipe($out_r, $out_w) or fail($self, "pipe failed: $!"); - if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ - fcntl($out_w, 1031, 4096); - fcntl($in_w, 1031, 4096) if $batch eq '--batch-check'; - } - my @cmd = (qw(git), "--git-dir=$self->{git_dir}", qw(-c core.abbrev=40 cat-file), $batch); - my $redir = { 0 => fileno($out_r), 1 => fileno($in_w) }; + my $redir = { 0 => $out_r }; if ($err) { - open(my $fh, '+>', undef) or fail($self, "open.err failed: $!"); + my $id = "git.$self->{git_dir}$batch.err"; + my $fh = tmpfile($id) or fail($self, "tmpfile($id): $!"); $self->{$err} = $fh; - $redir->{2} = fileno($fh); + $redir->{2} = $fh; } - my $p = spawn(\@cmd, undef, $redir); - defined $p or fail($self, "spawn failed: $!"); + my ($in_r, $p) = popen_rd(\@cmd, undef, $redir); $self->{$pid} = $p; $out_w->autoflush(1); + if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ + fcntl($out_w, 1031, 4096); + fcntl($in_r, 1031, 4096) if $batch eq '--batch-check'; + } $self->{$out} = $out_w; $self->{$in} = $in_r; } -sub cat_file { - my ($self, $obj, $ref) = @_; - my ($retried, $in, $head); - -again: - batch_prepare($self); - $self->{out}->print($obj, "\n") or fail($self, "write error: $!"); +sub my_read ($$$) { + my ($fh, $rbuf, $len) = @_; + my $left = $len - length($$rbuf); + my $r; + while ($left > 0) { + $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf)); + if ($r) { + $left -= $r; + } else { + next if (!defined($r) && $! == EINTR); + return $r; + } + } + \substr($$rbuf, 0, $len, ''); +} - $in = $self->{in}; - local $/ = "\n"; - $head = $in->getline; - if ($head =~ / missing$/) { - if (!$retried && alternates_changed($self)) { - $retried = 1; - cleanup($self); - goto again; +sub my_readline ($$) { + my ($fh, $rbuf) = @_; + while (1) { + if ((my $n = index($$rbuf, "\n")) >= 0) { + return substr($$rbuf, 0, $n + 1, ''); } - return; + my $r = sysread($fh, $$rbuf, $PIPE_BUFSIZ, length($$rbuf)); + next if $r || (!defined($r) && $! == EINTR); + return defined($r) ? '' : undef; # EOF or error } - $head =~ /^[0-9a-f]{40} \S+ (\d+)$/ or - fail($self, "Unexpected result from git cat-file: $head"); - - my $size = $1; - my $ref_type = $ref ? ref($ref) : ''; - - my $rv; - my $left = $size; - $$ref = $size if ($ref_type eq 'SCALAR'); - my $cb_err; - - if ($ref_type eq 'CODE') { - $rv = eval { $ref->($in, \$left) }; - $cb_err = $@; - # drain the rest - my $max = 8192; - while ($left > 0) { - my $r = read($in, my $x, $left > $max ? $max : $left); - defined($r) or fail($self, "read failed: $!"); - $r == 0 and fail($self, 'exited unexpectedly'); - $left -= $r; +} + +sub cat_async_retry ($$$$$) { + my ($self, $inflight, $req, $cb, $arg) = @_; + + # {inflight} may be non-existent, but if it isn't we delete it + # here to prevent cleanup() from waiting: + delete $self->{inflight}; + cleanup($self); + + $self->{inflight} = $inflight; + batch_prepare($self); + my $buf = "$req\n"; + for (my $i = 0; $i < @$inflight; $i += 3) { + $buf .= "$inflight->[$i]\n"; + } + print { $self->{out} } $buf or fail($self, "write error: $!"); + unshift(@$inflight, \$req, $cb, $arg); # \$ref to indicate retried + + cat_async_step($self, $inflight); # take one step +} + +sub cat_async_step ($$) { + my ($self, $inflight) = @_; + die 'BUG: inflight empty or odd' if scalar(@$inflight) < 3; + my ($req, $cb, $arg) = splice(@$inflight, 0, 3); + my $rbuf = delete($self->{cat_rbuf}) // \(my $new = ''); + my ($bref, $oid, $type, $size); + my $head = my_readline($self->{in}, $rbuf); + if ($head =~ /^([0-9a-f]{40}) (\S+) ([0-9]+)$/) { + ($oid, $type, $size) = ($1, $2, $3 + 0); + $bref = my_read($self->{in}, $rbuf, $size + 1) or + fail($self, defined($bref) ? 'read EOF' : "read: $!"); + chop($$bref) eq "\n" or fail($self, 'LF missing after blob'); + } elsif ($head =~ / missing$/) { + # ref($req) indicates it's already been retried + if (!ref($req) && !$in_cleanup && alternates_changed($self)) { + return cat_async_retry($self, $inflight, + $req, $cb, $arg); } + $type = 'missing'; + $oid = ref($req) ? $$req : $req; } else { - my $offset = 0; - my $buf = ''; - while ($left > 0) { - my $r = read($in, $buf, $left, $offset); - defined($r) or fail($self, "read failed: $!"); - $r == 0 and fail($self, 'exited unexpectedly'); - $left -= $r; - $offset += $r; - } - $rv = \$buf; + fail($self, "Unexpected result from async git cat-file: $head"); } + eval { $cb->($bref, $oid, $type, $size, $arg) }; + $self->{cat_rbuf} = $rbuf if $$rbuf ne ''; + warn "E: $oid: $@\n" if $@; +} - my $r = read($in, my $buf, 1); - defined($r) or fail($self, "read failed: $!"); - fail($self, 'newline missing after blob') if ($r != 1 || $buf ne "\n"); - die $cb_err if $cb_err; +sub cat_async_wait ($) { + my ($self) = @_; + my $inflight = delete $self->{inflight} or return; + while (scalar(@$inflight)) { + cat_async_step($self, $inflight); + } +} - $rv; +sub batch_prepare ($) { + _bidi_pipe($_[0], qw(--batch in out pid)); } -sub batch_prepare ($) { _bidi_pipe($_[0], qw(--batch in out pid)) } +sub _cat_file_cb { + my ($bref, undef, undef, $size, $result) = @_; + @$result = ($bref, $size); +} -sub check { - my ($self, $obj) = @_; - _bidi_pipe($self, qw(--batch-check in_c out_c pid_c err_c)); - $self->{out_c}->print($obj, "\n") or fail($self, "write error: $!"); - local $/ = "\n"; - chomp(my $line = $self->{in_c}->getline); - my ($hex, $type, $size) = split(' ', $line); +sub cat_file { + my ($self, $oid, $sizeref) = @_; + my $result = []; + cat_async($self, $oid, \&_cat_file_cb, $result); + cat_async_wait($self); + $$sizeref = $result->[1] if $sizeref; + $result->[0]; +} - # Future versions of git.git may show 'ambiguous', but for now, +sub check_async_step ($$) { + my ($self, $inflight_c) = @_; + die 'BUG: inflight empty or odd' if scalar(@$inflight_c) < 3; + my ($req, $cb, $arg) = splice(@$inflight_c, 0, 3); + my $rbuf = delete($self->{rbuf_c}) // \(my $new = ''); + chomp(my $line = my_readline($self->{in_c}, $rbuf)); + my ($hex, $type, $size) = split(/ /, $line); + + # Future versions of git.git may have type=ambiguous, but for now, # we must handle 'dangling' below (and maybe some other oddball # stuff): # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/ - return if $type eq 'missing' || $type eq 'ambiguous'; - if ($hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop') { - $size = $type + length("\n"); - my $r = read($self->{in_c}, my $buf, $size); - defined($r) or fail($self, "read failed: $!"); - return; + my $ret = my_read($self->{in_c}, $rbuf, $type + 1); + fail($self, defined($ret) ? 'read EOF' : "read: $!") if !$ret; } + eval { $cb->($hex, $type, $size, $arg, $self) }; + warn "E: check($req) $@\n" if $@; + $self->{rbuf_c} = $rbuf if $$rbuf ne ''; +} +sub check_async_wait ($) { + my ($self) = @_; + my $inflight_c = delete $self->{inflight_c} or return; + while (scalar(@$inflight_c)) { + check_async_step($self, $inflight_c); + } +} + +sub check_async_begin ($) { + my ($self) = @_; + cleanup($self) if alternates_changed($self); + _bidi_pipe($self, qw(--batch-check in_c out_c pid_c err_c)); + die 'BUG: already in async check' if $self->{inflight_c}; + $self->{inflight_c} = []; +} + +sub check_async ($$$$) { + my ($self, $oid, $cb, $arg) = @_; + my $inflight_c = $self->{inflight_c} // check_async_begin($self); + if (scalar(@$inflight_c) >= MAX_INFLIGHT) { + check_async_step($self, $inflight_c); + } + print { $self->{out_c} } $oid, "\n" or fail($self, "write error: $!"); + push(@$inflight_c, $oid, $cb, $arg); +} + +sub _check_cb { # check_async callback + my ($hex, $type, $size, $result) = @_; + @$result = ($hex, $type, $size); +} + +sub check { + my ($self, $oid) = @_; + my $result = []; + check_async($self, $oid, \&_check_cb, $result); + check_async_wait($self); + my ($hex, $type, $size) = @$result; + + # Future versions of git.git may show 'ambiguous', but for now, + # we must handle 'dangling' below (and maybe some other oddball + # stuff): + # https://public-inbox.org/git/20190118033845.s2vlrb3wd3m2jfzu@dcvr/T/ + return if $type eq 'missing' || $type eq 'ambiguous'; + return if $hex eq 'dangling' || $hex eq 'notdir' || $hex eq 'loop'; ($hex, $type, $size); } sub _destroy { - my ($self, $in, $out, $pid, $expire) = @_; - my $rfh = $self->{$in} or return; - if (defined $expire) { - # at least FreeBSD 11.2 and Linux 4.20 update mtime of the - # read end of a pipe when the pipe is written to; dunno - # about other OSes. - my $mtime = (stat($rfh))[9]; - return if $mtime > $expire; - } + my ($self, $rbuf, $in, $out, $pid, $err) = @_; my $p = delete $self->{$pid} or return; - foreach my $f ($in, $out) { - delete $self->{$f}; - } - waitpid $p, 0; + delete @$self{($rbuf, $in, $out)}; + delete $self->{$err} if $err; # `err_c' + + # PublicInbox::DS may not be loaded + eval { PublicInbox::DS::dwaitpid($p, undef, undef) }; + waitpid($p, 0) if $@; # wait synchronously if not in event loop +} + +sub cat_async_abort ($) { + my ($self) = @_; + my $inflight = delete $self->{inflight} or die 'BUG: not in async'; + cleanup($self); } sub fail { my ($self, $msg) = @_; - cleanup($self); - die $msg; + $self->{inflight} ? cat_async_abort($self) : cleanup($self); + croak("git $self->{git_dir}: $msg"); } sub popen { @@ -237,7 +332,6 @@ sub popen { sub qx { my ($self, @cmd) = @_; my $fh = $self->popen(@cmd); - defined $fh or return; local $/ = "\n"; return <$fh> if wantarray; local $/; @@ -246,9 +340,13 @@ sub qx { # returns true if there are pending "git cat-file" processes sub cleanup { - my ($self, $expire) = @_; - _destroy($self, qw(in out pid), $expire); - _destroy($self, qw(in_c out_c pid_c), $expire); + my ($self) = @_; + local $in_cleanup = 1; + delete $self->{async_cat}; + check_async_wait($self); + cat_async_wait($self); + _destroy($self, qw(cat_rbuf in out pid)); + _destroy($self, qw(chk_rbuf in_c out_c pid_c err_c)); !!($self->{pid} || $self->{pid_c}); } @@ -259,7 +357,7 @@ sub packed_bytes { my ($self) = @_; my $n = 0; my $pack_dir = git_path($self, 'objects/pack'); - foreach my $p (glob("$pack_dir/*.pack")) { + foreach my $p (bsd_glob("$pack_dir/*.pack", GLOB_NOSORT)) { $n += -s $p; } $n @@ -277,25 +375,82 @@ sub local_nick ($) { wantarray ? ($ret) : $ret; } -# show the blob URL for cgit/gitweb/whatever -sub src_blob_url { - my ($self, $oid) = @_; - # blob_url_format = "https://example.com/foo.git/blob/%s" - if (my $bfu = $self->{blob_url_format}) { - return map { sprintf($_, $oid) } @$bfu if wantarray; - return sprintf($bfu->[0], $oid); - } - local_nick($self); +sub host_prefix_url ($$) { + my ($env, $url) = @_; + return $url if index($url, '//') >= 0; + my $scheme = $env->{'psgi.url_scheme'}; + my $host_port = $env->{HTTP_HOST} // + "$env->{SERVER_NAME}:$env->{SERVER_PORT}"; + "$scheme://$host_port". ($env->{SCRIPT_NAME} || '/') . $url; } sub pub_urls { - my ($self) = @_; + my ($self, $env) = @_; if (my $urls = $self->{cgit_url}) { - return @$urls; + return map { host_prefix_url($env, $_) } @$urls; } local_nick($self); } +sub cat_async_begin { + my ($self) = @_; + cleanup($self) if alternates_changed($self); + batch_prepare($self); + die 'BUG: already in async' if $self->{inflight}; + $self->{inflight} = []; +} + +sub cat_async ($$$;$) { + my ($self, $oid, $cb, $arg) = @_; + my $inflight = $self->{inflight} // cat_async_begin($self); + if (scalar(@$inflight) >= MAX_INFLIGHT) { + cat_async_step($self, $inflight); + } + + print { $self->{out} } $oid, "\n" or fail($self, "write error: $!"); + push(@$inflight, $oid, $cb, $arg); +} + +# this is safe to call inside $cb, but not guaranteed to enqueue +# returns true if successful, undef if not. +sub async_prefetch { + my ($self, $oid, $cb, $arg) = @_; + if (defined($self->{async_cat}) && (my $inflight = $self->{inflight})) { + # we could use MAX_INFLIGHT here w/o the halving, + # but lets not allow one client to monopolize a git process + if (scalar(@$inflight) < int(MAX_INFLIGHT/2)) { + print { $self->{out} } $oid, "\n" or + fail($self, "write error: $!"); + return push(@$inflight, $oid, $cb, $arg); + } + } + undef; +} + +sub extract_cmt_time { + my ($bref, undef, undef, undef, $modified) = @_; + + if ($$bref =~ /^committer .*?> ([0-9]+) [\+\-]?[0-9]+/sm) { + my $cmt_time = $1 + 0; + $$modified = $cmt_time if $cmt_time > $$modified; + } +} + +# returns the modified time of a git repo, same as the "modified" field +# of a grokmirror manifest +sub modified ($) { + my ($self) = @_; + my $modified = 0; + my $fh = popen($self, qw(rev-parse --branches)); + local $/ = "\n"; + while (my $oid = <$fh>) { + chomp $oid; + cat_async($self, $oid, \&extract_cmt_time, \$modified); + } + cat_async_wait($self); + $modified || time; +} + 1; __END__ =pod