This is caused by pthread_cancel effectively throwing a
not-specifically-identifiable C++ exception into the targeted thread,
which, if it is not rethrown, terminates the process entirely.
This is rather "impolite" behaviour, we would say. But thread
cancellation is *always* busted, and we should simply not use it where
unnecessary. It's particularly unnecessary when what we *actually* need
it for is, err, interrupting a poll(2).
That can in turn be achieved by simply listening to more stuff in the
poll, namely, a pipe, which we send a character to when needing to
stop the thread.
While looking at this code, we also investigated whether any of the
poll() madness is required, or was even *ever* required. Curiously we
found in the XNU kernel source code that the thing about needing to
listen to POLLHUP is probably *correct*, but switching it to POLLRDNORM
should not have made any difference at all. We've left a FIXME to look
into that further because what's written here is super janky.
94d3b45284/bsd/kern/sys_generic.c (L1751-L1758)
This is the crash on some Hydra machines:
Thread 1 (Thread 0x7f56b77776c0 (LWP 955542) (Exiting)):
0 0x00007f56b8e9b7dc in __pthread_kill_implementation () from /nix/store/m71p7f0nymb19yn1dascklyya2i96jfw-glibc-2.39-52/lib/libc.so.6
1 0x00007f56b8e49516 in raise () from /nix/store/m71p7f0nymb19yn1dascklyya2i96jfw-glibc-2.39-52/lib/libc.so.6
2 0x00007f56b8e31935 in abort () from /nix/store/m71p7f0nymb19yn1dascklyya2i96jfw-glibc-2.39-52/lib/libc.so.6
3 0x00007f56b8e327f3 in __libc_message_impl.cold () from /nix/store/m71p7f0nymb19yn1dascklyya2i96jfw-glibc-2.39-52/lib/libc.so.6
4 0x00007f56b8e8e8e9 in __libc_fatal () from /nix/store/m71p7f0nymb19yn1dascklyya2i96jfw-glibc-2.39-52/lib/libc.so.6
5 0x00007f56b8ea23c4 in unwind_cleanup () from /nix/store/m71p7f0nymb19yn1dascklyya2i96jfw-glibc-2.39-52/lib/libc.so.6
6 0x00007f56b9d2a1b8 in nix::triggerInterrupt() [clone .cold] () from /nix/store/sahgw550p621m9dy1pd7whl9c5g1g0p7-lix-2.90.0-rc1/lib/liblixutil.so
7 0x00007f56b990ac9d in std:🧵:_State_impl<std:🧵:_Invoker<std::tuple<nix::MonitorFdHup::MonitorFdHup(int)::{lambda()#1}> > >::_M_run() () from /nix/store/sahgw550p621m9dy1pd7whl9c5g1g0p7-lix-2.90.0-rc1/lib/liblixstore.so
8 0x00007f56b90e86d3 in execute_native_thread_routine () from /nix/store/c6r62m84hywf4i6qq1h28f13zv38yqyp-gcc-13.3.0-lib/lib/libstdc++.so.6
9 0x00007f56b8e99a42 in start_thread () from /nix/store/m71p7f0nymb19yn1dascklyya2i96jfw-glibc-2.39-52/lib/libc.so.6
10 0x00007f56b8f1905c in clone3 () from /nix/store/m71p7f0nymb19yn1dascklyya2i96jfw-glibc-2.39-52/lib/libc.so.6
As for testing, we've started a daemon with this change and verified it
deals with HUPs correctly on x86_64-linux, but I don't think we can
easily test the destructor behaviour without whatever Hydra was
doing that broke.
Change-Id: I29c7de0425674494b6e43c075810126c3ff77363
This also bans various sneaking of negative numbers from the language
into unsuspecting builtins as was exposed while auditing the
consequences of changing the Nix language integer type to a newtype.
It's unlikely that this change comprehensively ensures correctness when
passing integers out of the Nix language and we should probably add a
checked-narrowing function or something similar, but that's out of scope
for the immediate change.
During the development of this I found a few fun facts about the
language:
- You could overflow integers by converting from unsigned JSON values.
- You could overflow unsigned integers by converting negative numbers
into them when going into Nix config, into fetchTree, and into flake
inputs.
The flake inputs and Nix config cannot actually be tested properly
since they both ban thunks, however, we put in checks anyway because
it's possible these could somehow be used to do such shenanigans some
other way.
Note that Lix has banned Nix language integer overflows since the very
first public beta, but threw a SIGILL about them because we run with
-fsanitize=signed-overflow -fsanitize-undefined-trap-on-error in
production builds. Since the Nix language uses signed integers, overflow
was simply undefined behaviour, and since we defined that to trap, it
did.
Trapping on it was a bad UX, but we didn't even entirely notice
that we had done this at all until it was reported as a bug a couple of
months later (which is, to be fair, that flag working as intended), and
it's got enough production time that, aside from code that is IMHO buggy
(and which is, in any case, not in nixpkgs) such as
lix-project/lix#445, we don't think
anyone doing anything reasonable actually depends on wrapping overflow.
Even for weird use cases such as doing funny bit crimes, it doesn't make
sense IMO to have wrapping behaviour, since two's complement arithmetic
overflow behaviour is so *aggressively* not what you want for *any* kind
of mathematics/algorithms. The Nix language exists for package
management, a domain where bit crimes are already only dubiously in
scope to begin with, and it makes a lot more sense for that domain for
the integers to never lose precision, either by throwing errors if they
would, or by being arbitrary-precision.
This change will be ported to CppNix as well, to maintain language
consistency.
Fixes: lix-project/lix#423
Change-Id: I51f253840c4af2ea5422b8a420aa5fafbf8fae75
The actual motive here is the avoidance of integer overflow if we were
to make these use checked NixInts and retain the subtraction.
However, the actual *intent* of this code is a three-way comparison,
which can be done with operator<=>, so we should just do *that* instead.
Change-Id: I7f9a7da1f3176424b528af6d1b4f1591e4ab26bf
upcast_goal was only ever needed to break circular includes, but the
same solution that gave us upcast_goal also lets us fully remove it:
just upcast goals without a wrapper function, but only in .cc files.
Change-Id: I9c71654b2535121459ba7dcfd6c5da5606904032
this will let us turn copyNAR into a generator as well, which in turn is
necessary to turn the users of copyNAR into generators without resorting
to sinkToSource coroutines. currently this uses the SerializingTransform
in all cases, even for copyNAR where it is not necessary. should this be
a performance problem we can easily swap out the transform for one which
does not produce any bytes of its own, but that should not be necessary.
Change-Id: I7e685879318fcbb78d8b88abfddd7752360eb0ce
the sole remaining user of this function can use makeDecompressionSource
instead, while making the sinkToSource in the caller unnecessary as well
Change-Id: I4258227b5dbbb735a75b477d8a57007bfca305e9
the rewriting sink was just broken. when given a rewrite set that
contained a key that is also a proper infix of another key it was
possible to produce an incorrectly rewritten result if the writer
used the wrong block size. fixing this duplicates rewriteStrings,
to avoid this we'll rewrite rewriteStrings to use RewritingSource
in a new mode that'll allow rewrites we had previously forbidden.
Change-Id: I57fa0a9a994e654e11d07172b8e31d15f0b7e8c0
This updates the version of rnix used and refactors the code generally
to be more precise and capable in it's identification of both lambdas
and determining which documentation comments are attached.
Change-Id: Ib0dddabd71f772c95077f9d7654023b37a7a1fd2
this is supposed to be a set of outputs we want to always succeed for
releases. sadly we can't add nixos installer tests using lix to these
because the nixos test framework does not allow overriding nix in the
installer test suites due to unfortunate oversights in the framework.
Change-Id: I815520181ccca70a47205d38ba27e73529347f04
we want to be sure we can cross-build to aarch64 for releases, add a
target to our crossSystems list to make those cheacks easier to run.
Change-Id: Ieb65c1333a5232641ace0ba4d122fc7d528ebc04
`nix-collect-garbage --dry-run` previously elided the entire garbage
collection check, meaning that it would just exit the script without
printing anything.
This change makes the dry run flag instead set the GC action to
`gcReturnDead` rather than `gcDeleteDead`, and then continue with the
script. So if you set `--dry-run`, it will print the paths it *would*
have garbage collected, but not actually delete them.
I filed a bug for this: lix-project/lix#432 but then realised I could give fixing it a go myself.
Change-Id: I062dbf1a80bbab192b5fd0b3a453a0b555ad16f2
DrvInfo's query methods all use mutable fields to cache, but like.
that's basically the entire interface for DrvInfo. Not only that, but
these formerly-const-marked functions can even throw due to eval errors!
Changing this only required removing some `const` markers in nix-env,
and changing a single inline `queryInstalled()` call to be an lvalue
instead.
Change-Id: I796807118f3b35b0e93668b5e28210d9e521b2ae
Activities can set display attributes in their log output using the "Select
Graphics Rendition" functionality. To prevent interfering with subsequent text
displayed, these should be reset after writing the log line. The multiline
progress bar neglected to do this, resulting for example in a colorised
"building …" header in the next line. Reset the attributes properly, like the
standard progress bar already does.
Change-Id: I1dc69f4a1d747a76b83e8721a72d9bb0e5554488
This rather simple function existed just to check some flags,
but the response varies by platform. This is a perfect case for
our subclasses.
Change-Id: Ieb1732a8d024019236e0d0028ad843a24ec3dc59
size tracking can be done with a LengthSink and a tee. match tracking
was defeated by never having done any match tracking, all users would
see the same (empty) set of matches at all times. match tracking with
bytes offsets alone would not be sufficient in the general case, only
because computeHashModulo uses a single rewrite could it have worked.
Change-Id: Idb214b5222e0ea24f450f5505712a342b63d7570
this much more closely mimics what is actually happening: we're reading
data from somewhere else, actively, rather than passively waiting. with
the data flow matching the underlying system interactions better we can
remove a few sinkToSource calls that merely exists to undo the mismatch
caused by not treating subprocess output as a data source to begin with
Change-Id: If4abfc2f8398fb5e88c9b91a8bdefd5504bb2d11
this will let us also return a source for the program output later,
which will in turn make sinkToSource unnecessary for program output
processing. this may also reopen a path for provigin program input,
but that still needs a proper async io framework to avoid problems.
Change-Id: Iaf93f47db99c38cfaf134bd60ed6a804d7ddf688
Turns errors like this:
let
throwMsg = a: throw (a + " invalid bar");
in throwMsg "bullshit"
error:
… from call site
at «string»:3:4:
2| throwMsg = a: throw (a + " invalid bar");
3| in throwMsg "bullshit"
| ^
… while calling 'throwMsg'
at «string»:2:14:
1| let
2| throwMsg = a: throw (a + " invalid bar");
| ^
3| in throwMsg "bullshit"
… while calling the 'throw' builtin
at «string»:2:17:
1| let
2| throwMsg = a: throw (a + " invalid bar");
| ^
3| in throwMsg "bullshit"
error: bullshit invalid bar
into errors like this:
let
throwMsg = a: throw (a + " invalid bar");
in throwMsg "bullshit"
error:
… from call site
at «string»:3:4:
2| throwMsg = a: throw (a + " invalid bar");
3| in throwMsg "bullshit"
| ^
… while calling 'throwMsg'
at «string»:2:14:
1| let
2| throwMsg = a: throw (a + " invalid bar");
| ^
3| in throwMsg "bullshit"
… caused by explicit throw
at «string»:2:17:
1| let
2| throwMsg = a: throw (a + " invalid bar");
| ^
3| in throwMsg "bullshit"
error: bullshit invalid bar
Change-Id: I593688928ece20f97999d1bf03b2b46d9ac338cb
Turns errors like:
let
somepkg.src = throw "invalid foobar";
in somepkg.src.meta
error:
… while evaluating the attribute 'src.meta'
at «string»:2:3:
1| let
2| somepkg.src = throw "invalid foobar";
| ^
3| in somepkg.src.meta
… while calling the 'throw' builtin
at «string»:2:17:
1| let
2| somepkg.src = throw "invalid foobar";
| ^
3| in somepkg.src.meta
error: invalid foobar
into errors like:
let
somepkg.src = throw "invalid foobar";
in somepkg.src.meta
error:
… while evaluating the attribute 'src.meta'
at «string»:2:3:
1| let
2| somepkg.src = throw "invalid foobar";
| ^
3| in somepkg.src.meta
… while evaluating 'somepkg.src' to select 'meta' on it
at «string»:3:4:
2| somepkg.src = throw "invalid foobar";
3| in somepkg.src.meta
| ^
… while calling the 'throw' builtin
at «string»:2:17:
1| let
2| somepkg.src = throw "invalid foobar";
| ^
3| in somepkg.src.meta
error: invalid foobar
And for type errors, from:
let
somepkg.src = "I'm not an attrset";
in somepkg.src.meta
error:
… while evaluating the attribute 'src.meta'
at «string»:2:3:
1| let
2| somepkg.src = "I'm not an attrset";
| ^
3| in somepkg.src.meta
… while selecting an attribute
at «string»:3:4:
2| somepkg.src = "I'm not an attrset";
3| in somepkg.src.meta
| ^
error: expected a set but found a string: "I'm not an attrset"
into:
let
somepkg.src = "I'm not an attrset";
in somepkg.src.meta
error:
… while evaluating the attribute 'src.meta'
at «string»:2:3:
1| let
2| somepkg.src = "I'm not an attrset";
| ^
3| in somepkg.src.meta
… while selecting 'meta' on 'somepkg.src'
at «string»:3:4:
2| somepkg.src = "I'm not an attrset";
3| in somepkg.src.meta
| ^
error: expected a set but found a string: "I'm not an attrset"
For the low price of an enumerate() and a lambda you too can have the
incorrect line of code actually show up in the trace!
Change-Id: Ic1491c86e33c167891bdac9adad6224784760bd6
Turns errors like:
let
errpkg = throw "invalid foobar";
in errpkg.meta
error:
… while calling the 'throw' builtin
at «string»:2:12:
1| let
2| errpkg = throw "invalid foobar";
| ^
3| in errpkg.meta
error: invalid foobar
into errors like:
let
errpkg = throw "invalid foobar";
in errpkg.meta
error:
… while evaluating 'errpkg' to select 'meta' on it
at «string»:3:4:
2| errpkg = throw "invalid foobar";
3| in errpkg.meta
| ^
… while calling the 'throw' builtin
at «string»:2:12:
1| let
2| errpkg = throw "invalid foobar";
| ^
3| in errpkg.meta
error: invalid foobar
For the low price of one try/catch, you too can have the incorrect line
of code actually show up in the trace!
Change-Id: If8d6200ec1567706669d405c34adcd7e2d2cd29d
Add a platform-specific function for starting sandboxed child.
Generally this just means startProcess, but on Linux we use flags
for clone to start a new namespace
Change-Id: I41c8aba62676a162388bbe5ab8a7518904c7b058
Add a new OS-specific hook called `prepareSandbox`, run before forking
On Darwin this is empty as nothing is required,
on Linux this creates the chroot directory and adds basic files,
and on platforms using a fallback this throws an exception
Change-Id: Ie30c38c387f2e0e5844b2afa32fd4d33b1180dae
generators are a better basis for serializers than streaming into sinks
as we do currently for many reasons, such as being usable as sources if
one wishes to (without requiring an intermediate sink to serialize full
data sets into memory, or boost coroutines to turn sinks into sources),
composing more naturally (as one can just yield a sub-generator instead
of being forced to wrap entire substreams into clunky functions or even
more clunky custom types to implement operator<< on), allowing wrappers
to transform data with clear ownership semantics (removing the need for
explicit memory allocations and Source wrappers), and many other things
Change-Id: I361d89ff556354f6930d9204f55117565f2f7f20
the `*Source` name is a slight misnomer since we do also have a
Source type, but we can probably live with this for time being.
Change-Id: I54eb2e59a4009014e324797f16b80b962759c7d3
not used anywhere yet, but we'll use this a lot soon for generators that
return file contents, wire protocol fragments, or indeed any byte stream
Change-Id: I01a46f9bf9d75aaf4a5d7662773b99f498862a28
this will be the basis of non-boost coroutines in lix. anything that is
a boost coroutine *should* be representable with a Generator coroutine,
and many things that are not currently boost coroutines but behave much
like one (such as, notably, serializers) should be as well. this allows
us to greatly simplify many things that look like iteration but aren't.
Change-Id: I2cebcefa0148b631fb30df4c8cfa92167a407e34
not printing activities at all when no progress information is available
hides *all* progress information from e.g. flake show. this is not ideal
and needs to be fixed, but the fix *still* has problems with flake show:
in multiline mode we will overwrite all useful flake show output as soon
as the progress bar is redrawn. flake show output is also mangled in any
number of other situations (like -v being set), so we should probably be
not too worried about it and fix progress reporting properly another day
Change-Id: I6d39d670e261bbae00560b6a8e15dec8e16b35c4
Previously, the progress bar had two subtly different states in which the bar
would not actually render, both with their own shortcomings: inactive (which
was irreversible) and paused (reversible, but swallowing logs). Furthermore,
there was no way of resetting the statistics, so a very bad solution was
implemented (243c0f18da) that would create a new
logger for each line of the repl, leaking the previous one and discarding the
value of printBuildLogs. Finally, if stderr was not attached to a TTY, the
update thread was started even though the logger was not active, violating the
invariant required by the destructor (which is not observed because the logger
is leaked).
In this commit, the two aforementioned states are unified into a single one,
which can be exited again, correctly upholds the invariant that the update
thread is only running while the progress bar is active, and does not swallow
logs. The latter change in behavior is not expected to be a problems in the
rare cases where the paused state was used before, since other loggers (like
the simple one) don't exhibit it anyway. The startProgressBar/stopProgressBar
API is removed due to being a footgun, and a new method for properly resetting
the progress is added.
Co-Authored-By: Qyriad <qyriad@qyriad.me>
Change-Id: I2b7c3eb17d439cd0c16f7b896cfb61239ac7ff3a
jade
Lunaphied
eldritch horrors
piegames
alois31
Quantum Jump
Qyriad
Artemis Tosini
Delan Azabani