Commit graph

9950 commits

Author SHA1 Message Date
Eelco Dolstra
70bcd6a55c Evaluation cache: Don't barf in read-only mode
Fixes

  $ nix copy
  warning: Git tree '/home/eelco/Dev/nix-flake' is dirty
  nix: src/nix/installables.cc:348: std::tuple<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, nix::FlakeRef, nix::InstallableValue::DerivationInfo> nix::InstallableFlake::toDerivation(): Assertion `state->store->isValidPath(drvPath)' failed.
  Aborted (core dumped)
2020-04-29 15:42:53 +02:00
Guillaume Bouchard
5a34a473dd builtins.readFile: do not truncate content
This closes #3026 by allowing `builtins.readFile` to read a file with a
wrongly reported file size, for example, files in `/proc` may report a
file size of 0. Reading file in `/proc` is not a good enough motivation,
however I do think it just makes nix more robust by allowing more file
to be read.  Especially, I do considerer the previous behavior to be
dangerous because nix was previously reading truncated files. Examples
of file system which incorrectly report file size may be network file
system or dynamic file system (for performance reason, a dynamic file
system such as FUSE may generate the content of the file on demand).

```
nix-repl> builtins.readFile "/proc/version"
""
```

With this commit:

```
nix-repl> builtins.readFile "/proc/version"
"Linux version 5.6.7 (nixbld@localhost) (gcc version 9.3.0 (GCC)) #1-NixOS SMP Thu Apr 23 08:38:27 UTC 2020\n"
```

Here is a summary of the behavior changes:

- If the reported size is smaller, previous implementation
was silently returning a truncated file content. The new implementation
is returning the correct file content.

- If a file had a bigger reported file size, previous implementation was
failing with an exception, but the new implementation is returning the
correct file content. This change of behavior is coherent with this pull
request.

Open questions

- The behavior is unchanged for correctly reported file size, however
performances may vary because it uses the more complex sink interface.
Considering that sink is used a lot, I don't think this impacts the
performance a lot.
- `builtins.readFile` on an infinite file, such as `/dev/random` may
fill the memory.
- it does not support adding file to store, such as `${/proc/version}`.
2020-04-29 14:50:52 +02:00
Eelco Dolstra
5ada0831cf Merge remote-tracking branch 'origin/master' into flakes 2020-04-29 13:32:27 +02:00
Ben Burdette
22e6490311 Error classname as name 2020-04-28 21:06:08 -06:00
Eelco Dolstra
06849c3090
Merge pull request #3542 from mkenigs/gcroots
Set GCROOT to store path to prevent garbage collection
2020-04-28 21:04:06 +02:00
Matthew Kenigsberg
6d40fe573c rename to NIX_GCROOT 2020-04-28 11:18:54 -06:00
Eelco Dolstra
52a3ca823d Tweak warning message 2020-04-28 17:56:01 +02:00
Eelco Dolstra
6a8cba83bb Merge branch 'nix-env-warn-unmatched' of https://github.com/lheckemann/nix 2020-04-28 17:45:25 +02:00
Eelco Dolstra
ee754f0f41
Merge pull request #3541 from alyssais/gcdos
Fix long paths permanently breaking GC
2020-04-28 16:33:45 +02:00
Ben Burdette
e51a757720 astyle format 2020-04-27 15:15:08 -06:00
Eelco Dolstra
6521c92ce8 Improve path:// handling
In particular, doing 'nix build /path/to/dir' now works if
/path/to/dir is not a Git tree (it only has to contain a flake.nix
file).

Also, 'nix flake init' no longer requires a Git tree (but it will do a
'git add flake.nix' if it's a Git tree)
2020-04-27 22:53:11 +02:00
Eelco Dolstra
829dcb35d5 flake-template.nix: Add defaultPackage 2020-04-27 22:52:49 +02:00
Alyssa Ross
c05e20daa1
Fix long paths permanently breaking GC
Suppose I have a path /nix/store/[hash]-[name]/a/a/a/a/a/[...]/a,
long enough that everything after "/nix/store/" is longer than 4096
(MAX_PATH) bytes.

Nix will happily allow such a path to be inserted into the store,
because it doesn't look at all the nested structure.  It just cares
about the /nix/store/[hash]-[name] part.  But, when the path is deleted,
we encounter a problem.  Nix will move the path to /nix/store/trash, but
then when it's trying to recursively delete the trash directory, it will
at some point try to unlink
/nix/store/trash/[hash]-[name]/a/a/a/a/a/[...]/a.  This will fail,
because the path is too long.  After this has failed, any store deletion
operation will never work again, because Nix needs to delete the trash
directory before recreating it to move new things to it.  (I assume this
is because otherwise a path being deleted could already exist in the
trash, and then moving it would fail.)

This means that if I can trick somebody into just fetching a tarball
containing a path of the right length, they won't be able to delete
store paths or garbage collect ever again, until the offending path is
manually removed from /nix/store/trash.  (And even fixing this manually
is quite difficult if you don't understand the issue, because the
absolute path that Nix says it failed to remove is also too long for
rm(1).)

This patch fixes the issue by making Nix's recursive delete operation
use unlinkat(2).  This function takes a relative path and a directory
file descriptor.  We ensure that the relative path is always just the
name of the directory entry, and therefore its length will never exceed
255 bytes.  This means that it will never even come close to AX_PATH,
and Nix will therefore be able to handle removing arbitrarily deep
directory hierachies.

Since the directory file descriptor is used for recursion after being
used in readDirectory, I made a variant of readDirectory that takes an
already open directory stream, to avoid the directory being opened
multiple times.  As we have seen from this issue, the less we have to
interact with paths, the better, and so it's good to reuse file
descriptors where possible.

I left _deletePath as succeeding even if the parent directory doesn't
exist, even though that feels wrong to me, because without that early
return, the linux-sandbox test failed.

Reported-by: Alyssa Ross <hi@alyssa.is>
Thanks-to: Puck Meerburg <puck@puckipedia.com>
Tested-by: Puck Meerburg <puck@puckipedia.com>
Reviewed-by: Puck Meerburg <puck@puckipedia.com>
2020-04-27 20:50:17 +00:00
Matthew Kenigsberg
9e95b95a5d comment 2020-04-27 13:18:26 -06:00
Matthew Kenigsberg
a3bc695e7d Set GCROOT to store path to prevent garbage collection 2020-04-27 11:22:20 -06:00
Eelco Dolstra
b51dff431c Improve error message when an argument is not a flake 2020-04-27 18:55:20 +02:00
Eelco Dolstra
b4e23dcd9e nix search: Search legacyPackages recursively 2020-04-27 16:29:26 +02:00
Ben Burdette
1ff42722ce error.hh 2020-04-26 14:47:41 -06:00
Ben Burdette
d4fd7b543e print dashes instead of empty name string 2020-04-25 12:05:26 -06:00
Linus Heckemann
f59404e1a6 nix-env: refactor uninstallDerivations
Reduces the number of store queries it performs. Also prints a warning
if any of the selectors did not match any installed derivations.

UX Caveats:
- Will print a warning that nothing matched if a previous selector
  already removed the path
- Will not do anything if no selectors were provided (no change from
  before).

Fixes #3531
2020-04-25 16:30:42 +02:00
Ben Burdette
cdac083dc5 don't print blank lines for blank description 2020-04-24 21:40:13 -06:00
Ben Burdette
d8d4844b88 all things error to error.hh 2020-04-24 14:57:51 -06:00
Ben Burdette
d9632765a8 add has_value check; remove obslete friend class 2020-04-24 12:44:23 -06:00
Eelco Dolstra
0469795978 nix search: Show version 2020-04-24 14:42:17 +02:00
Ben Burdette
833501f6f1 'what' string 2020-04-23 15:55:34 -06:00
Eelco Dolstra
ef4d3fc111 Merge remote-tracking branch 'origin/master' into flakes 2020-04-23 15:16:18 +02:00
Ben Burdette
3bc9155dfc a few more 'format's rremoved 2020-04-22 15:00:11 -06:00
Eelco Dolstra
c9d0cf7e02
Don't include error.hh in util.hh to prevent header bloat 2020-04-22 15:29:27 +02:00
Eelco Dolstra
7114f088fc
Don't install error-demo 2020-04-22 15:29:22 +02:00
Eelco Dolstra
16e3bf4537
Merge branch 'error-format' of https://github.com/bburdette/nix 2020-04-22 15:29:10 +02:00
Eelco Dolstra
2ea4d45449
Path fetcher: Fix store path name
(cherry picked from commit c7af247bea)
2020-04-22 15:27:06 +02:00
Eelco Dolstra
4a2a45f53d
Merge pull request #3522 from HackerFoo/replace-select-with-poll
Replace select() with poll()
2020-04-22 12:25:41 +02:00
Eelco Dolstra
c7af247bea Path fetcher: Fix store path name 2020-04-22 10:15:32 +02:00
Dustin DeWeese
c0d940978a Replace select() with poll() to allow waiting on more than FD_SETSIZE fds 2020-04-21 16:21:28 -07:00
Ben Burdette
e4fb9a3849 remove 'format' from Error constructor calls 2020-04-21 17:07:07 -06:00
Ben Burdette
d3052197fe add ErrorInfo to BaseError 2020-04-21 13:25:41 -06:00
Eelco Dolstra
8c75621da6 Fix typo 2020-04-20 15:28:56 +02:00
Eelco Dolstra
b69323f8c9 Revive 'nix search'
It uses the evaluation cache now rather than the ad hoc JSON cache.
2020-04-20 15:27:09 +02:00
Eelco Dolstra
42a12f9232 Move eval-cache.{cc,hh} 2020-04-20 13:14:59 +02:00
Eelco Dolstra
539a9c1c5f Get rid of the old eval cache 2020-04-20 13:13:52 +02:00
Ben Burdette
15e9564fd1 logEI for tunnelLogger and progressbar 2020-04-19 17:16:51 -06:00
Eelco Dolstra
0725ab2fd7 Store more stuff in the evaluation cache
In particular, we store whether an attribute failed to evaluate (threw
an exception) or was an unsupported type. This is to ensure that a
repeated 'nix flake show' never has to evaluate anything, so it can
execute without fetching the flake.

With this, 'nix flake show nixpkgs/nixos-20.03 --legacy' executes in
0.6s (was 3.4s).
2020-04-19 23:07:06 +02:00
Eelco Dolstra
3738bcb05e Eval cache: Don't replace real attributes with placeholders 2020-04-18 15:12:31 +02:00
Domen Kožar
25ed842725
Merge pull request #3502 from NixOS/more-pos
pass Pos to forceValue to improve infinite recursion error
2020-04-18 14:05:21 +02:00
Ben Burdette
4697552948 demoing other error levels than warn/error; rename line and file fields in errPos 2020-04-17 15:50:46 -06:00
Eelco Dolstra
69cb9f7eee Wrap eval cache creation in a giant transaction
This speeds up the creation of the cache for the nixpkgs flake from
21.2s to 10.2s. Oddly, it also speeds up querying the cache
(i.e. running 'nix flake show nixpkgs/nixos-20.03 --legacy') from 4.2s
to 3.4s.

(For comparison, running with --no-eval-cache takes 9.5s, so the
overhead of building the SQLite cache is only 0.7s.)
2020-04-17 23:17:21 +02:00
Ben Burdette
3d5b1032a1 logError, logWarning; Logger functions; switch to Verbosity enum 2020-04-17 15:07:44 -06:00
Eelco Dolstra
aaa109565e Use a more space/time-efficient representation for the eval cache 2020-04-17 23:04:21 +02:00
Eelco Dolstra
bdb3226607 Add flag to disable the eval cache 2020-04-17 14:30:04 +02:00
Eelco Dolstra
aa34c0ef51 nix flake show: Speed up eval cache bigly
In the fully cached case for the 'nixpkgs' flake, it went from 101s to
4.6s. Populating the cache went from 132s to 17.4s (which could
probably be improved further by combining INSERTs).
2020-04-17 13:57:02 +02:00