As of GDB 13.1, gdb.exe contains static data that triggers a bug in x64
Windows zip integration as far back as Windows 7. After recompressing
with "advzip -z4", extracting gdb.exe fails with "Unspecified error"
0x80004005. It happens with both 32-bit and 64-bit gdb.exe builds, and
regardless of compiler flags, which is why I'm confident it's static
data. No other zip program I've tried has trouble with this file.
WIP NOTE: I thought it might not happen with -z3, but it still does. The
committed "advzip.sample" contains the bad pattern.
Programs with main typically also have mainCRTStartup, but it's unlikely
that the user wants to break inside the CRT. If WinMain or wWinMain are
present, the user purposefully defined them, so prioritize those even
higher.
Windows programs have several conventional entry points depending on
various circumstances. Upstream GDB is only aware of "main" and the
"start" command behaves poorly or incorrectly otherwise.
I've gotten a lot better at CRT-free since I first wrote this. I'm more
confident about this version, it produces better error messages, and it
compiles around 10x faster. The last point matters because w64devkit
currently compiles this program 181 times.
This is the first official versioned release of Universal Ctags, dated
December 2022. There were no prior tagged releases, let alone stable
source tarballs, so w64devkit piggy-backed off Debian's fork for the
needed stability.
This release has an official, stable source tarball, but it's broken and
practically useless. Further, the Git repository isn't tagged correctly,
having no annotated tags even for 6.0.0. Fortunately GitHub has recently
committed to generating stable source tarballs following backlash from
archive breakage in early February 2023. Because GitHub's source tarball
is more reliable than the official tarball, use it instead.
As of GDB 12.1, libiconv is a mandatory dependency on Windows. As with
Expat, this fact is undocumented. GDB builds successfully without but
that build is broken and practically useless. This dependency introduces
no new features nor makes GDB work better (i.e. GDB still cannot display
non-ASCII strings). It is necessary only to retain the capabilities
already present in GDB 11 an earlier.
This introduces yet another dependency, but it's a substantial gain for
a small cost.
GDB lacks "--with-libcurses-prefix" so I had to manually list the path
in LDFLAGS. Otherwise the GDB configure script fails to find waddstr and
gives up even though actually linking gdb.exe works fine without it.
The capitalized file name for PDCurses introduces a small sorting issue
in SHA256SUMS. Lexicographic collation with PDCurses at the top? A more
natural collation with PDCurses in the middle? I generate the listing
via "sha256sum *" in Bash, and in my locale Bash uses the latter, so
that's what I'll use.
Obviously elfedit and readelf are not useful on Windows. GProf has not
worked on Windows since Binutils 2.36 because it doesn't support PIE. It
was never effective in the first place, so it probably wasn't ever worth
including.
In the future I plan to call these "CRT-free" rather than "freestanding"
since it is more accurate. This approach has been trial-and-error due to
its unusual nature, the lack of convention, non-existing documentation,
and poor implicit defaults in the GNU toolchain. After spending time
navigating the problem space and discovering its pitfalls, I understand
much better how to construct these binaries. The changes at a high level
and their reasoning:
Avoid -ffreestanding since this is not really its purpose. It may still
be necessary to use -fno-builtin to prevent certain CRT calls.
Replace -fno-ident with -Wl,--gc-sections. It accomplishes the same and
much more. Even better, it is only needed at link time.
Do not qualify mainCRTStartup with WINAPI. By default, the GNU linker
looks for this symbol as the entry point. Not finding one, it silently
chooses the beginning of the .text section as the entry point. Marked
WINAPI, it gets __stdcall-decorated on i686 and is no longer recognized
by the linker. In the past I have been lucky that "main" happened to be
the first function in .text, so this worked out. This is incompatible
with the MSVC toolchain (an oversight in MinGW / Mingw-w64) since it is
explicitly documented that the entry point uses __stdcall. Since there
are no arguments, __stdcall and __cdecl happen to be identical, which is
why nobody notices.
Alternatively I could use -Wl,--entry= to choose an entry point, which
at least produces a warning when the symbol is not found. However, this
accepts the decorated symbol name, and so the exact invocation depends
on the target: i686 or x86-64. Rather inconvenient. The implicit search
for mainCRTStartup resolves this automatically, except when __stdcall is
involved, per above.
If using more than two pages of local variables in any function, you
must disable the stack probe with -mno-stack-arg-probe because the probe
is provided by the CRT. Normally the stack grows with use, and the probe
handles large stack frames that may jump the guard page. Instead commit
a larger stack at startup: "-Xlinker --stack=Z,Z" where Z is the desired
size, e.g. use 0x100000 to commit a 1MiB stack.
The one-shot command works, but only use a single core. This build
benefits significantly from parallelization, and make is the easiest way
to do it. Since the original Cppcheck makefile doesn't do the right
thing, supply a custom build.
Even the latest GDB release does not correctly support DWARF 5 from GCC,
most notably preprocessor macros. This is a bug in either GCC or GDB and
is present across every platform tested. DWARF 4 works fine, so stick to
it for now.
This program is compiled many times, and 99% of the build time is GCC
parsing windows.h repeatedly. This macro trims the header and cuts the
build time almost in half, which will improve the overall w64devkit
build time.
