This seems to be the root cause of the random page allocation failures
and @wizeman did a very good job on not only finding the root problem
but also giving a detailed explanation of it in #10828.
Here is an excerpt:
The problem here is that the kernel is trying to allocate a contiguous
section of 2^7=128 pages, which is 512 KB. This is way too much:
kernel pages tend to get fragmented over time and kernel developers
often go to great lengths to try allocating at most only 1 contiguous
page at a time whenever they can.
From the error message, it looks like the culprit is unionfs, but this
is misleading: unionfs is the name of the userspace process that was
running when the system ran out of memory, but it wasn't unionfs who
was allocating the memory: it was the kernel; specifically it was the
v9fs_dir_readdir_dotl() function, which is the code for handling the
readdir() function in the 9p filesystem (the filesystem that is used
to share a directory structure between a qemu host and its VM).
If you look at the code, here's what it's doing at the moment it tries
to allocate memory:
buflen = fid->clnt->msize - P9_IOHDRSZ;
rdir = v9fs_alloc_rdir_buf(file, buflen);
If you look into v9fs_alloc_rdir_buf(), you will see that it will try
to allocate a contiguous buffer of memory (using kzalloc(), which is a
wrapper around kmalloc()) of size buflen + 8 bytes or so.
So in reality, this code actually allocates a buffer of size
proportional to fid->clnt->msize. What is this msize? If you follow
the definition of the structures, you will see that it's the
negotiated buffer transfer size between 9p client and 9p server. On
the client side, it can be controlled with the msize mount option.
What this all means is that, the reason for running out of memory is
that the code (which we can't easily change) tries to allocate a
contiguous buffer of size more or less equal to "negotiated 9p
protocol buffer size", which seems to be way too big (in our NixOS
tests, at least).
After that initial finding, @lethalman tested the gnome3 gdm test
without setting the msize parameter at all and it seems to have resolved
the problem.
The reason why I'm committing this without testing against all of the
NixOS VM test is basically that I think we can only go better but not
worse than the current state.
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
Previously is was assumed that bash was in the path when calling the
environment setup script. This changes all of the references of bash to
be absolute paths so that the user doesn't have to worry about the
environment they call it with.
Otherwise, if the upstream mirror changes (rather than deletes) a
file, then tarballs.nixos.org won't be used even if it has a copy of
the original file, and so we'll get a hash mismatch.
Emacs packages are commonly distributed as single .el files. This
unpackCmd handles them correctly and sets up sourceRoot. Other sources
are treated in the default manner.
If "fetcher" is a string, then Nix will execute it with bash already, so
the additional bash argument in that string was redundant and apparently
causes trouble on non-Linux platforms.
Hopefully fixes https://github.com/NixOS/nixpkgs/issues/11496.
The list we had before contained a lot of junk, i.e. sites that were no
longer online or no longer in sync. The new list of sites comes from
https://gnupg.org/download/index.html.
It turns out that cargo implicitly depends on rustc at runtime: even
`cargo help` will fail if rustc is not in the PATH.
This means that we need to wrap the cargo binary to add rustc to PATH.
However, I have opted into doing something slightly unusual: instead of
tying down a specific cargo to use a specific rustc (i.e., wrap cargo so
that "${rustc}/bin" is prefixed into PATH), instead I'm adding the rustc
used to build cargo as a fallback rust compiler (i.e., wrap cargo so
that "${rustc}/bin" is suffixed into PATH). This means that cargo will
prefer to use a rust compiler that is in the default path, but fallback
into the one used to build cargo only if there wasn't any rust compiler
in the default path.
The reason I'm doing this is that otherwise it could cause unexpected
effects. For example, if you had a build environment with the
rustcMaster and cargo derivations, you would expect cargo to use
rustcMaster to compile your project (since rustcMaster would be the only
compiler available in $PATH), but this wouldn't happen if we tied down
cargo to use the rustc that was used to compile it (because the default
cargo derivation gets compiled with the stable rust compiler).
That said, I have slightly modified makeRustPlatform so that a rust
platform will always use the rust compiler that was used to build cargo,
because this prevents mistakenly depending on two different versions of
the rust compiler (stable and unstable) in the same rust platform,
something which is usually undesirable.
Fixes #11053
