git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: Optimizing writes to unchanged files during merges?

From
PWPhillip Wood <phillip.wood@talktalk.net>
Date
Apr 16, 2018, 17:47 UTC
Message-ID
<21d15f6b-f4ab-2efc-47a9-6cbf95cf80d7@talktalk.net>
In-Reply-To
<F1738316-71EF-4053-82E5-F009F491CCE8@gmail.com>
On 16/04/18 17:07, Lars Schneider wrote:
Show 27 quoted lines
> 
> I am happy to see this discussion and the patches, because long rebuilds 
> are a constant annoyance for us. We might have been bitten by the exact 
> case discussed here, but more often, we have a slightly different 
> situation:
> 
> An engineer works on a task branch and runs incremental builds — all 
> is good. The engineer switches to another branch to review another 
> engineer's work. This other branch changes a low-level header file, 
> but no rebuild is triggered. The engineer switches back to the previous 
> task branch. At this point, the incremental build will rebuild 
> everything, as the compiler thinks that the low-level header file has
> been changed (because the mtime is different).
> 
> Of course, this problem can be solved with a separate worktree. However, 
> our engineers forget about that sometimes, and then, they are annoyed by 
> a 4h rebuild.
> 
> Is this situation a problem for others too?
> If yes, what do you think about the following approach:
> 
> What if Git kept a LRU list that contains file path, content hash, and 
> mtime of any file that is removed or modified during a checkout. If a 
> file is checked out later with the exact same path and content hash, 
> then Git could set the mtime to the previous value. This way the 
> compiler would not think that the content has been changed since the 
> last rebuild.
Hi Lars

But if there has been rebuild between the checkouts then you want the compiler to rebuild. I've been using the script below recently to save and restore mtimes around running rebase to squash fixup commits. To avoid restoring the mtimes if there has been a rebuild since they were stored it takes a list of build sentinels and stores their mtimes too - if any of those change then it will refuse to restore the original mtimes of the tracked files (if you give a path that does exist when the mtimes are stored then it will refuse to restore the mtimes if that path exists when you run 'git mtimes restore'). The sentinels can be specified on the commandline when running 'git mtimes save' or stored in multiple mtimes.sentinal config keys.

Best Wishes
Phillip
--->8---
#!/usr/bin/perl

# Copyright (C) 2018 Phillip Wood <phillip.wood@dunelm.org.uk> # # git-mtimes.perl # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2 of the License, or # (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, see <http://www.gnu.org/licenses/>.

use 5.008; use strict; use warnings; use File::Copy (qw(copy)); use File::Spec::Functions (qw(abs2rel catfile file_name_is_absolute rel2abs)); use Storable ();

sub git;

my $GIT_DIR = git(qw(rev-parse --git-dir)) or exit 1; $GIT_DIR = rel2abs($GIT_DIR); my $mtimes_path = "$GIT_DIR/mtimes";

sub git {
    my @lines;
    # in a scalar context slurp removing any trailing $/
    # in an array context return a list of lines
    {
	local $/ = wantarray ? $/ : undef;
	local $,=' ';
	open my $fh, '-|', 'git', @_ or die "git @_ failed $!";
	@lines = <$fh>;
	chomp @lines;
	unless (close $fh) {
	    $? == -1 and die "git @_ not found";
	    my $code = $? >> 8;
	    $_[0] eq 'config' and $code == 1 or
		die "git @_ failed with exit code $code"
	}
    }
    wantarray and return @lines;
    @lines and chomp @lines;
    return $lines[0];
}
sub ls_files {
    # mode, uid, gid, mtime and maybe atime
    my @stat_indices = $_[0] ? (2, 4, 5, 9, 8) : (2, 4, 5, 9);
    local $_;
    local $/ = "\0";
    my @files;
    for (git(qw(ls-files --stage -z))) {
	if (/^[^ ]+ ([^\t]+) 0\t(.*)/) {
	    my @stat = stat($2);
	    # store name, hash, mode, uid, gid, mtime and maybe atime
	    push @files, [ $2, $1, @stat[@stat_indices] ];
	}
    }
    return @files;
}
sub get_config {
    local $/ = "\0";
    my $get = wantarray ? '--get-all' : '--get';
    git(qw(config -z), $get, @_);
}
sub save {
    local $_;
    my @sentinels = get_config('mtimes.sentinel');
    push @sentinels, @ARGV;
    @sentinels or die "No sentinels given";
    @sentinels = map { [ $_, [ stat $_ ] ] } @sentinels;
    my @files = ls_files();
    Storable::nstore( [ [ @sentinels ] , [ @files ] ], $mtimes_path) or
	die "unable to store mtimes $!";
}
sub match_sentinel_data {
    local $_;
    my ($old, $new, $trustctime) = @_;
    if (!@$old) {
	return (@$new) ? undef : 1;
    } else {
	@$new or return undef;
    }
    # Skip hardlink count, atime, blksize
    for (0..2,4..7,9,10,12) {
	next if ($_ == 10 and ! $trustctime);
	$old->[$_] == $new->[$_] or return undef;
    }
    return 1;
}
sub needs_update {
    local $_;
    my ($old, $new) = @_;
    for (0..1) {
	$old->[$_] eq $new->[$_] or return undef;
    }
    for (2..4) {
	$old->[$_] == $new->[$_] or return undef;
    }
    $old->[5] != $new->[5];
}
sub restore {
    local $_;
    my $stored = Storable::retrieve($mtimes_path) or
	die "unable to load stored data";
    my $trustctime = get_config('--bool', 'core.trustctime');
    $trustctime = defined($trustctime) ? $trustctime eq 'true' : 1;
    my ($sentinels, $oldfiles) = @$stored;
    for (@$sentinels) {
	match_sentinel_data( [ stat($_->[0]) ], $_->[1], $trustctime) or
	    die "Unable to restore mtimes, stat data for sentinel '$_->[0]' does not match";
    }
    my @newfiles = ls_files(1);
    my ($i, $restored) = (0, 0);
    for (@$oldfiles) {
	while ($newfiles[$i]->[0] lt $_->[0] and $i < @newfiles) {
	    $i++;
	}
	if (needs_update($_, $newfiles[$i])) {
	    utime($newfiles[$i]->[6], $_->[5], $_->[0]);
	    $restored = 1;
	}
    }
    if ($restored) {
	print "restored mtimes\n";
    }
}
my $cmd = shift;
# Keep relative paths relative in case repository directory is renamed
# between saving and restoring mtimes.
if ($ENV{GIT_PREFIX}) {
    @ARGV = map {
	file_name_is_absolute($_) ? $_ : catfile($ENV{GIT_PREFIX}, $_);
    } @ARGV;
}
my $up = git(qw(rev-parse --show-cdup));
if ($up) {
    @ARGV = map {
	file_name_is_absolute($_) ? $_ : abs2rel(rel2abs($_), $up);
    } @ARGV;
    chdir $up;
}
my $tmp_index = catfile($GIT_DIR, "mtimes-index");
my $src_index = $ENV{GIT_INDEX_FILE} ? $ENV{GIT_INDEX_FILE} :
				       catfile($GIT_DIR, "index");
copy($src_index, $tmp_index) or
    die "cannot create temporary index '$tmp_index'\n";
$ENV{GIT_INDEX_FILE} = $tmp_index;
git(qw(add -u));
if ($cmd eq 'save') {
    save();
} elsif ($cmd eq 'restore' and ! @ARGV) {
    restore();
} else {
    print STDERR "usage: git mtimes <save [sentinels ...] | restore>\n";
}
END {
    unlink $tmp_index;
}
Show 6 quoted lines
> I think that would fix the problem that our engineers run into and also 
> the problem that Linus experienced during the merge, wouldn't it?
> 
> Thanks,
> Lars
> 
Previous: Jacob KellerNext: Stefan Haller
Message 24 of 29 in “Optimizing writes to unchanged files during merges?”
  1. Linus TorvaldsApr 12, 2018
  2. Junio C HamanoApr 12, 2018
  3. Junio C HamanoApr 12, 2018
  4. Linus TorvaldsApr 12, 2018
  5. Linus TorvaldsApr 12, 2018
  6. Linus TorvaldsApr 12, 2018
  7. Linus TorvaldsApr 13, 2018
  8. Elijah NewrenApr 13, 2018
  9. Linus TorvaldsApr 13, 2018
  10. Stefan BellerApr 13, 2018
  11. Linus TorvaldsApr 13, 2018
  12. Elijah NewrenApr 13, 2018
  13. Junio C HamanoApr 13, 2018
  14. Junio C HamanoApr 16, 2018
  15. Linus TorvaldsApr 16, 2018
  16. Lars SchneiderApr 16, 2018
  17. Ævar Arnfjörð BjarmasonApr 16, 2018
  18. Lars SchneiderApr 17, 2018
  19. Jacob KellerApr 16, 2018
  20. Jacob KellerApr 16, 2018
  21. Junio C HamanoApr 16, 2018
  22. Lars SchneiderApr 17, 2018
  23. Jacob KellerApr 17, 2018
  24. Phillip WoodApr 16, 2018
  25. Stefan HallerApr 16, 2018
  26. Elijah NewrenApr 16, 2018
  27. Elijah NewrenApr 16, 2018
  28. Linus TorvaldsApr 12, 2018
  29. Elijah NewrenApr 13, 2018

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.