We now have a C++ red-black tree implementation that implements standard
template library compatible APIs while compiling 10x faster than libcxx.
It's not as beautiful as the red-black tree implementation in Plinko but
this will get the job done and the test proves it upholds all invariants
This change also restores CheckForMemoryLeaks() support and fixes a real
actual bug I discovered with Doug Lea's dlmalloc_inspect_all() function.
It hasn't been helpful enough to be justify the maintenance burden. What
actually does help is mprotect(), kprintf(), --ftrace and --strace which
can always be counted upon to work correctly. We aren't losing much with
this change. Support for ASAN on AARCH64 was never implemented. Applying
ASAN to the core libc runtimes was disabled many months ago. If there is
some way to have an ASAN runtime for user programs that is less invasive
we can potentially consider reintroducing support. But now is premature.
Actually Portable Executable now supports Android. Cosmo's old mmap code
required a 47 bit address space. The new implementation is very agnostic
and supports both smaller address spaces (e.g. embedded) and even modern
56-bit PML5T paging for x86 which finally came true on Zen4 Threadripper
Cosmopolitan no longer requires UNIX systems to observe the Windows 64kb
granularity; i.e. sysconf(_SC_PAGE_SIZE) will now report the host native
page size. This fixes a longstanding POSIX conformance issue, concerning
file mappings that overlap the end of file. Other aspects of conformance
have been improved too, such as the subtleties of address assignment and
and the various subtleties surrounding MAP_FIXED and MAP_FIXED_NOREPLACE
On Windows, mappings larger than 100 megabytes won't be broken down into
thousands of independent 64kb mappings. Support for MAP_STACK is removed
by this change; please use NewCosmoStack() instead.
Stack overflow avoidance is now being implemented using the POSIX thread
APIs. Please use GetStackBottom() and GetStackAddr(), instead of the old
error-prone GetStackAddr() and HaveStackMemory() APIs which are removed.
* Add ctl utility.h
Implements forward, move, swap, and declval. This commit also adds a def
for nullptr_t to cxx.inc. We need it now because the CTL headers stopped
including anything from libc++, so we no longer get their basic types.
* Use ctl::swap in string
The STL spec says that swap is located in the string_view header anyawy.
Performance-wise this is a noop, but it’s slightly cleaner.
c.inc (AFAICT erroneously) defined _Atomic(t) as `volatile t *`, when it
should have just said `volatile t`, when __STDC_VERSION__ was too small.
This happens when we’re compiling C++, but in C++11, _Atomic is a define
supplied by the STL rather than a keyword supplied by the compiler. Wait
though, it gets better: in C++11, _Atomic hooks you into the morass that
is stdatomic.h, and ultimately refers everything back to std::atomic<T>.
The gory, horrifying details are in libcxx's __atomic/cxx_atomic_impl.h.
The tldr is that for our purposes it’s fine to just say volatile and use
the normal libc/intrin/atomic.h functions.
🚨 clang-format changes output per version!
This is with version 19.0.0. The modifications seem to be fixing the old
version’s errors - mainly involving omitted whitespace around binary ops
and inserted whitespace between goto labels and colons (if followed by a
curly brace.)
Also fixes a few mistakes made by e.g. someone (ahem) forgetting to pass
his ctl/string.h modifications through it.
We should add this to .git-blame-ignore-revs once we have its final hash
on master.
If pthread_create() is linked into the binary, then the cosmo runtime
will create an independent dlmalloc arena for each core. Whenever the
malloc() function is used it will index `g_heaps[sched_getcpu() / 2]`
to find the arena with the greatest hyperthread / numa locality. This
may be configured via an environment variable. For example if you say
`export COSMOPOLITAN_HEAP_COUNT=1` then you can restore the old ways.
Your process may be configured to have anywhere between 1 - 128 heaps
We need this revision because it makes multithreaded C++ applications
faster. For example, an HTTP server I'm working on that makes extreme
use of the STL went from 16k to 2000k requests per second, after this
change was made. To understand why, try out the malloc_test benchmark
which calls malloc() + realloc() in a loop across many threads, which
sees a a 250x improvement in process clock time and 200x on wall time
The tradeoff is this adds ~25ns of latency to individual malloc calls
compared to MODE=tiny, once the cosmo runtime has transitioned into a
fully multi-threaded state. If you don't need malloc() to be scalable
then cosmo provides many options for you. For starters the heap count
variable above can be set to put the process back in single heap mode
plus you can go even faster still, if you include tinymalloc.inc like
many of the programs in tool/build/.. are already doing since that'll
shave tens of kb off your binary footprint too. Theres also MODE=tiny
which is configured to use just 1 plain old dlmalloc arena by default
Another tradeoff is we need more memory now (except in MODE=tiny), to
track the provenance of memory allocation. This is so allocations can
be freely shared across threads, and because OSes can reschedule code
to different CPUs at any time.
Cosmo will now print C++ symbols correctly in --ftrace logs and
backtraces. Doing this required reducing the memory requirement
of the __demangle() function by 3x. This was accomplished using
16-bit indices and 16-bit malloc granularity. That puts a limit
on the longest symbol we can successfully decode, which I think
would be around 6553 characters long, given a 65536-byte buffer
We're now able to rewind the instruction pointer in x86 backtraces. This
helps ensure addr2line cannot print information about unrelated adjacent
code. I've restored -fno-schedule-insns2 in most cases because it really
does cause unpredictable breakage for backtraces.
This change fixes a bug where exiting a crash signal handler on Windows
after adding the signal to uc_sigmask, but not correcting the CPU state
would cause the signal handler to loop infinitely, causing process hang
Another issue is that very tiny programs, that don't link posix signals
would not have their SIGILL / SIGSEGV / etc. status reported to Cosmo's
bash shell when terminating on crash. That's fixed by a tiny handler in
WinMain() that knows how to map WIN32 crash codes to the POSIX flavors.
Microsoft caused some very gentle breakages for Cosmopolitan. They
removed the version information from the PEB which caused uname to
report WINDOWS 0.0.0. We should have called GetVersionExW but that
doesn't really exist anymore either. Windows policy is now to give
whatever version we used in ape/ape.S. Windows8 has been EOL since
2023-01-10 so lets avoid our modern executables being relegated to
legacy infrastructure. Requiring Windows 10+ going forward lets us
remove runtime compatibility bloat from the codebase. Further note
Cosmopolitan maintains a Windows Vista branch on GitHub, so anyone
preferring the older versions, can still have a future with Cosmo.
Another neat thing this fixes is UTF-8 support in the console. The
changes Microsoft made broke the if statement that enabled UTF8 in
terminals. This explains why bug reports had broken arrows. In the
future this should be less of an issue, since the PEB code is gone
which means we more strictly conform to only Microsoft's WIN32 API
The normal getopt() function is bloated because it links printf(). This
change exports the original authentic bsd getopt function, that cosmo's
always used internally so cosmocc users don't need to include internals
This change adds a TLS freelist for small dynamic memory allocations.
Cosmopolitan's TIB is now 512 bytes in size. Single-threaded malloc()
performance isn't impacted by this, until pthread_create() is called.
Single-threaded programs may also want to consider using:
#include "libc/mem/tinymalloc.inc"
Which will shave 30k off the executable size and sometimes go faster.
This change solves the XNU crash loop mystery. Apple's documentation
claims to support this feature, but they only define the constant in
their header files. The kernel acknowledges thi SA_RESETHAND bit, by
clearing it from the sa_flags state, returns zero, and does nothing.
We're now able to pretty print a C++ backtrace upon crashing in pretty
much any runtime execution scenario. The default pledge sandbox policy
on Linux is now to return EPERM. If you call pledge and have debugging
functions linked (e.g. GetSymbolTable) then the symbol table shall get
loaded before any security policy is put in place. This change updates
build/bootstrap/fixupobj too and fixes some other sneaky build errors.
For this to work, a loader has to be able to tell the difference between
an ‘old’ and a ‘new’ binary. This is achieved via a repurposing of ELF’s
e_flags field. We previously tried to use the padding in e_ident for it,
but binutils was resetting it to zero in e.g. strip.
This introduces one new ELF flag for cosmopolitan binaries. It is called
`EF_APE_MODERN`. We choose 0x101ca75, "lol cat 5".
It should now be safe to install the ape loader binfmt registration with
the `P` flag.
It's now possible to safely print C++ backtraces from signal handlers.
This symbol demangler doesn't need malloc, tls, or even static memory.
Additionally, this change makes it 2x faster and adds test cases. It's
almost as performant and accurate as the libcxxabi implementation now.
Cosmopolitan now supports 104 time zones. They're embedded inside any
binary that links the localtime() function. Doing so adds about 100kb
to the binary size. This change also gets time zones working properly
on Windows for the first time. It's not needed to have /etc/localtime
exist on Windows, since we can get this information from WIN32. We're
also now updated to the latest version of Paul Eggert's TZ library.
The feenableexcept() and fedisableexcept() APIs are now provided which
let you detect when NaNs appear the moment it happens from anywhere in
your program. Tests have also been added for the mission critical math
functions expf() and erff(), whose perfect operation has been assured.
See examples/trapping.c to see how to use this powerful functionality.
* Fix fork locking on win32
- __enable_threads / set __threaded in __proc_setup as threads are required for
win32 subprocess management
- move mmi/fds locking out of pthread_atfork.c into fork.c so it's done anytime
__threaded is set instead of being dependent of pthreads
- explicitly yoink _pthread_onfork_prepare, _pthread_onfork_parent, and
_pthread_onfork_child in pthread_create.c so they are linked in in-case they
are separated from _pthread_atfork
Big Thanks to @dfyz for help with locating the issue, testing, and devising a fix!
* fix child processes not being able to open files, initialize all necessary locks on fork
Commit bc6c183 introduced a bunch of discrepancies between what files
look like in the repo and what clang-format says they should look like.
However, there were already a few discrepancies prior to that. Most of
these discrepancies seemed to be unintentional, but a few of them were
load-bearing (e.g., a #include that violated header ordering needing
something to have been #defined by a 'later' #include.)
I opted to take what I hope is a relatively smooth-brained approach: I
reverted the .clang-format change, ran clang-format on the whole repo,
reapplied the .clang-format change, reran clang-format again, and then
reverted the commit that contained the first run. Thus the full effect
of this PR should only be to apply the changed formatting rules to the
repo, and from skimming the results, this seems to be the case.
My work can be checked by applying the short, manual commits, and then
rerunning the command listed in the autogenerated commits (those whose
messages I have prefixed auto:) and seeing if your results agree.
It might be that the other diffs should be fixed at some point but I'm
leaving that aside for now.
fd '\.c(c|pp)?$' --print0| xargs -0 clang-format -i
The `com` parameter to `TryPath` was always 1, so there was no reason to
have it. This patch changes the logic to be as though `com` was 0, which
provides a possible answer to the TODO question -- the answer is no.
If we never care about appending `.com`, then `CopyWithCwd` doesn't need
to return anything beyond a boolean success value.
Previously, the atomic store looked like it was happening while the
struct's memory was still poisoned. I was unable to observe any issues
with this, but this change seems to make the code more obviously correct
(at the cost of a redundant atomic store to zeroed space in case the map
needed to be extended.)
Now that these functions are behind _COSMO_SOURCE there's no reason for
having the ugly underscore anymore. To use these functions, you need to
pass -mcosmo to cosmocc.
The WIN32 CreateProcess() function does not require an .exe or .com
suffix in order to spawn an executable. Now that we have Cosmo bash
we're no longer so dependent on the cmd.exe prompt.
- Write some more unit tests
- memcpy() on ARM is now faster
- Address the Musl complex math FIXME comments
- Some libm funcs like pow() now support setting errno
- Import the latest and greatest math functions from ARM
- Use more accurate atan2f() and log1pf() implementations
- atoi() and atol() will no longer saturate or clobber errno
Multiple projects I care about make the assumption that this isn't a
system call that sleeps for a particular number of nanonseconds, but
rather a function that parks processes on kernel scheduler quantums.
Anyone who wants the old behavior should use cosmo_clock_nanosleep()
* Fix reading the same symbol twice when using `{f,}scanf()`
PR #924 appears to use `unget()` subtly incorrectly when parsing
floating point numbers. The rest of the code only uses `unget()`
immediately followed by `goto Done;` to return back the symbol that
can't possibly belong to the directive we're processing.
With floating-point, however, the ungot characters could very well
be valid for the *next* directive, so we will essentially read them
twice. It can't be seen in `sscanf()` tests because `unget()` is a
no-op there, but the test I added for `fscanf()` fails like this:
...
EXPECT_EQ(0xDEAD, i1)
need 57005 (or 0xdead) =
got 908973 (or 0x000ddead)
...
EXPECT_EQ(0xBEEF, i2)
need 48879 (or 0xbeef) =
got 769775 (or 0x000bbeef)
This means we read 0xDDEAD instead of 0xDEAD and 0xBBEEF instead of
0xBEEF. I checked that both musl and glibc read 0xDEAD/0xBEEF, as
expected.
Fix the failing test by removing the unneeded `unget()` calls.
* Don't read invalid floating-point numbers in `*scanf()`
Currently, we just ignore any errors from `strtod()`. They can
happen either because no valid float can be parsed at all, or
because the state machine recognizes only a prefix of a valid
floating-point number.
Fix this by making sure `strtod()` parses everything we recognized,
provided it's non-empty. This requires to pop the last character
off the FP buffer, which is supposed to be parsed by the next
`*scanf()` directive.
* Make `%c` parsing in `*scanf()` respect the C standard
Currently, `%c`-style directives always succeed even if there
are actually fewer characters in the input than requested.
Before the fix, the added test fails like this:
...
EXPECT_EQ(2, sscanf("ab", "%c %c %c", &c2, &c3, &c4))
need 2 (or 0x02 or '\2' or ENOENT) =
got 3 (or 0x03 or '\3' or ESRCH)
...
EXPECT_EQ(0, sscanf("abcd", "%5c", s2))
need 0 (or 0x0 or '\0') =
got 1 (or 0x01 or '\1' or EPERM)
musl and glibc pass this test.
Sometimes we need to interact with code that wasn't compiled using
`-fno-omit-frame-pointer`. For example, if a function pointer gets
passed and called by a foreign function, linked by cosmo_dlopen().
Function call tracing will now detect backtrace pointer corruption
and simply reduce the indentation level back to zero, as a result.
This change upgrades to GCC 12.3 and GNU binutils 2.42. The GNU linker
appears to have changed things so that only a single de-duplicated str
table is present in the binary, and it gets placed wherever the linker
wants, regardless of what the linker script says. To cope with that we
need to stop using .ident to embed licenses. As such, this change does
significant work to revamp how third party licenses are defined in the
codebase, using `.section .notice,"aR",@progbits`.
This new GCC 12.3 toolchain has support for GNU indirect functions. It
lets us support __target_clones__ for the first time. This is used for
optimizing the performance of libc string functions such as strlen and
friends so far on x86, by ensuring AVX systems favor a second codepath
that uses VEX encoding. It shaves some latency off certain operations.
It's a useful feature to have for scientific computing for the reasons
explained by the test/libcxx/openmp_test.cc example which compiles for
fifteen different microarchitectures. Thanks to the upgrades, it's now
also possible to use newer instruction sets, such as AVX512FP16, VNNI.
Cosmo now uses the %gs register on x86 by default for TLS. Doing it is
helpful for any program that links `cosmo_dlopen()`. Such programs had
to recompile their binaries at startup to change the TLS instructions.
That's not great, since it means every page in the executable needs to
be faulted. The work of rewriting TLS-related x86 opcodes, is moved to
fixupobj.com instead. This is great news for MacOS x86 users, since we
previously needed to morph the binary every time for that platform but
now that's no longer necessary. The only platforms where we need fixup
of TLS x86 opcodes at runtime are now Windows, OpenBSD, and NetBSD. On
Windows we morph TLS to point deeper into the TIB, based on a TlsAlloc
assignment, and on OpenBSD/NetBSD we morph %gs back into %fs since the
kernels do not allow us to specify a value for the %gs register.
OpenBSD users are now required to use APE Loader to run Cosmo binaries
and assimilation is no longer possible. OpenBSD kernel needs to change
to allow programs to specify a value for the %gs register, or it needs
to stop marking executable pages loaded by the kernel as mimmutable().
This release fixes __constructor__, .ctor, .init_array, and lastly the
.preinit_array so they behave the exact same way as glibc.
We no longer use hex constants to define math.h symbols like M_PI.
- Introduce portable sched_getcpu() api
- Support GCC's __target_clones__ feature
- Make fma() go faster on x86 in default mode
- Remove some asan checks from core libraries
- WinMain() now ensures $HOME and $USER are defined
1. `libc/isystem/complex.h` (included when you do `#include <complex.h>`)
defines `_COMPLEX_H`, and then proceeds to include `libc/complex.h`,
which contains the actual complex-related declarations. However, they
are *also* guarded by `_COMPLEX_H` and hence effectively ignored.
Fix this by changing `_COMPLEX_H` to `COSMOPOLITAN_LIBC_COMPLEX_H_`,
which is consistent with what the other headers (such as `math.h`) do.
2. Cosmopolitan could only support IPv4 multicast requests for sockets,
since a declaration for `struct ipv6_mreq` was missing. Add support
for IPv6, too, by adding the missing declaration.
- Let OpenMP be usable via cosmocc
- Let libunwind be usable via cosmocc
- Make X86_HAVE(AVXVNNI) work correctly
- Avoid using MAP_GROWSDOWN on qemu-aarch64
- Introduce in6addr_any and in6addr_loopback
- Have thread stacks use MAP_GROWSDOWN by default
- Ask OpenMP to not use filesystem to manage threads
- Make NI_MAXHOST and NI_MAXSERV available w/o _GNU_SOURCE
We recently broke MODE=dbg support when we added C++ exception support.
This change adds the missing UBSAN interfaces, needed to get it working
again. Some of the ASAN checking in the SJLJ guts needed to be disabled
since I doubt anyone's combined the two features until now.
Embedding Blink builds in Cosmo executables was a failed experiment. It
turned out to be easier than expected to let the mono repo have support
for multiple architectures. Blink still works great; it's supported and
recommended; just please use it as a separate program. For example, you
can use Blink to run Cosmo binaries on architectures like i486 / s390x.
If you install qemu-user from apt then glibc links a lot of address
space bloat that causes pthread_create() to ENOMEM (a.k.a. EAGAIN).
Boosting the virtual memory quota from 512m to 2048m will hopefully
future proof the build for the future, as Linux distros get fatter.
Please note this only applies to MODE=aarch64 on x86_64 builds when
you're using QEMU from Debian/Ubuntu rather than installing the one
cosmo provides in third_party/qemu/qemu-aarch64.gz. This change may
also be useful to people who are using the host compiler toolchain.
- `__cxa_*` runtime functions are expected to be in the `abi` namespace,
which is currently an alias for `__cxxabiv1`.
- Rely on the header provided by `libcxxabi` for functions that we do
not implement ourselves anymore.
This will help C++ code that uses exceptions to be tinier. For example,
this change shaves away 1000 lines of assembly code from LLVM's libcxx,
which is 0.7% of all assembly instructions in the entire library.
We now store values in jmp_buf where the compiler wants them to be. This
fixes code that calls __builtin_setjmp() and __builtin_longjmp() such as
libunwind. All libcxxabi tests are now passing on ARM64.
See #1076
Renaming gc() to _gc() was a mistake since the better thing to do is put
it behind the _COSMO_SOURCE macro. We need this change because I haven't
wanted to use my amazing garbage collector ever since we renamed it. You
now need to define _COSMO_SOURCE yourself when using amalgamation header
and cosmocc users need to pass the -mcosmo flag to get the gc() function
Some other issues relating to cancelation have been fixed along the way.
We're also now putting cosmocc in a folder named `.cosmocc` so it can be
more safely excluded by grep --exclude-dir=.cosmocc --exclude-dir=o etc.
* third_party: Add libcxxabi
Added libcxxabi from LLVM 17.0.6
The library implements the Itanium C++ exception handling ABI.
* third_party/libcxxabi: Enable __cxa_thread_atexit
Enable `__cxa_thread_atexit` from libcxxabi.
`__cxa_thread_atexit_impl` is still implemented by the cosmo libc.
The original `__cxa_thread_atexit` has been removed.
* third_party/libcxx: Build with exceptions
Build libcxx with exceptions enabled.
- Removed `_LIBCPP_NO_EXCEPTIONS` from `__config`.
- Switched the exception implementation to `libcxxabi`. These two files
are taken from the same `libcxx` version as mentioned in `README.cosmo`.
- Removed `new_handler_fallback` in favor of `libcxxabi` implementation.
- Enable `-fexceptions` and `-frtti` for `libcxx`.
- Removed `THIRD_PARTY_LIBCXX` dependency from `libcxxabi` and
`libunwind`. These libraries do not use any runtime `libcxx` functions,
just headers.
* libc: Remove remaining redundant cxa functions
- `__cxa_pure_virtual` in `libcxxabi` is also a stub similar to the
existing one.
- `__cxa_guard_*` from `libcxxabi` is used instead of the ones from
Android.
Now there should be no more duplicate implementations.
`__cxa_thread_atexit_impl`, `__cxa_atexit`, and related supporting
functions, are still left to other libraries as in `libcxxabi`.
`libcxxabi` is also now added to `cosmopolitan.a` to make up for the
removed functions.
Affected in-tree libraries (`third_party/double-conversion`) have been
updated.
The AMD HIP SDK for Linux ships prebuilt DSOs with an RWX PT_GNU_STACK
since old versions of GCC made it nearly impossible to build artifacts
where that wasn't the case, however modern glibc systems will flat out
refuse to link RWX DSOs from an execuatble that uses PT_GNU_STACK = RW
The cosmo_dlsym() function now returns the raw function address. You
need to call cosmo_dltramp() on the result, to make it safe to call.
This change is important, because cosmo_dltramp() magic can't always
work; for some tricky functions, you need to translate ABIs by hand.
Now we do them for assimilated binaries (except on OpenBSD or XNU
non-Silicon), for XnuSilicon, and for binaries with the preserve-
argv[0] auxv flag set. We check whether to pass the argv[0] value
at the test site rather than the Child site. We move a lot of the
test initialization into Child in the non-child case, in order to
get at the pre-init value of `__program_executable_name`. Finally,
we print out info about what we are skipping.
If cosmo_dlopen() is linked on AMD64 then the runtime will switch to
using %gs for thread-local storage. This eliminates the need for the
imported symbol trampoline. It's now safer to pass function pointers
back and forth with imported libraries. Your program gets recompiled
at runtime to make it happen and the overhead is a few milliseconds.
The toolchain will now be downloaded going forward from multiple pinned
URLs which have shasums. Either wget or curl must be installed.
This change unblocks #1053
Our dynamic linking implementation is now able to support functions with
dozens of parameters. In addition to having extra integral arguments you
can now pass vector registers using intrinsic types. Lastly, you can now
return multiple values, which is useful for functions returning structs.
