In 02cb8503d2, I added the m_next and
m_nextpkt structures to mbuf, as per FreeBSD's mbufq system that
FreeBSD 11.1's net80211 code uses. What I didn't realize (and
korli and PulkoMandy who reviewed my code didn't notice either)
is that the data fields in mbuf are sometimes dealt with through
these LEN macros, which were now incorrect after such changes.
This caused an out-of-bounds memory write for data above a certain
size that was attempting to be written into an mbuf that under the
old sizing would have been fine, but under this new sizing was invalid,
and this manifested itself as a KDL under the guarded_heap (#14207).
It possibly also manifested itself as a stack-smash with the new net80211
code (uncommitted on a local machine, and the reason I tried using
the guarded heap in the first place.)
Now we use FreeBSD 11's macros, which use offsetof instead of raw
integer math. This means that we can't specify the struct size in mbuf
as these structs are computed from mbuf's definition, and thus have
to rely on the allocator giving mbuf the correct size (as FreeBSD
does also.)
FreeBSD's is presently 46 bytes. CID 1422869 warns that it can get overrun
in if_attach() in copying if_xname which is IF_NAMESIZE bytes (32).
This breaks ABI, but BeOS did not have sockaddr_dl, it is only a modern-GCC
ABI break. Since most applications assume that sockaddr_dl is variable-length
and is null-terminated, as well as not used very often, hopefully this will
require relatively few rebuilds.
No functional change. Ideally we would create these via ResAttr like we do
the mail-providers, but these have attributes on their directories and not
just the files, so that will be slightly trickier, so for now just keep them
in their zip format.
There is a (probably non-viable) case where it might not get initialized
by the function and then potentially used, so set it to NULL to silence
Clang.
Some users of MK_ERROR pass in parameters from a variable called "res",
which is obviously not what they want to do, as that will use this "res"
and not theirs.
Spotted by Clang.
* The if-case was appending to gccBaseFlags after the rest of the file
was done using it, so it was ineffective. Now we set it with the rest
of the baseFlags.
* We already pass no-integrated-as in configure, no need to do it in
MainBuildRules.
* B_USE_BUILTIN_ATOMIC_FUNCTIONS isn't used anymore, so get rid of it.
Now, instead of breaking them up, all settings related to or gleaned
from the compiler are listed first (except for BOOT_CXXFLAGS_...).
No functional change intended.
If one specifies a cross-tools path instead of --build-cross-tools along
with --use-clang, then the specified architecture winds up in the list twice,
as the first test looks for the arch name where only "unknown" exists
(since in the case of cross-tools paths, we delay fetching the arch.)
So we need to move this check to there instead.
This lets clang use our linker and other binutils instead of its own.
Now clang builds produce a working bootloader and get all the way
to the "rocket" icon, at which point userland init fails.
Most of the problems with tree-vrp stemmed from its deletion of null-pointer
checks (see linked commit in the source.) Now, GCC has a flag to control that,
and with it enabled I can boot to the desktop even with tree-vrp enabled.
* When Download window is initially shown, it is correctly
fully onscreen; however, adding a download resizes the
window such that it becomes partially offscreen. Now,
when the frame is resized, reposition the window onscreen
again.
Fixes#12704
The user iframe and associated data that the syscall entry pushes to the
stack directly were causing the stack to be mis-aligned by 8 bytes. Since
we re-aligned %rsp afterwards, for most usecases this wasn't a problem.
However, since we stored the pre-realinged %rsp in %rbp (as we need it to
access the iframe data), this also meant that anything which depended on
%rbp being 16-byte-aligned would run into serious problems.
As it turned out, GCC 7 assumed that %rbp was indeed 16-byte-aligned, and
so optimized certain accesses to use SSE instructions that depended on this
alignment. Since inside any callstack begining with a syscall this was not
the case, a "General Protection Exception" resulted (see #14160 for an example)
at the first usage of such an instruction. I wasn't really sure what was going
on when it first came up, and so "fixed" it by disabling the GCC optimization
that used such instructions. Replacing the -fdisable... with -mstackrealign thus
also "fixes" the problem, as I discovered earlier today, as it forces GCC to
realign the stack in function prologues.
So instead of rounding %rsp down to the nearest aligned address after the
pushes are complete, we offset %rsp by the amount the pushes are not,
thus fixing both %rsp and %rbp in syscall handling routines. This of course
depends on syscall_rsp being already aligned, which it is.
Thanks to PulkoMandy and js for the advice and guidance (and PulkoMandy
for the ASCII art), as this is essentially my first time working with
kernel assembly.
* Using attribute visibility hidden doesn't get applied if a
function returns a non-class pointer type, so the functions
weren't being hidden for gcc4+ builds, resulting in stack
overflows. Using addr_t, which should be the same size as
void* works around this restriction.
This reverts commit c558f9c8fe.
This reverts commit 44f24718b1.
This reverts commit a69cb33030.
This reverts commit 951182620e.
There have been multiple reports that these changes break mounting NTFS partitions
(on all systems, see #14204), and shutting down (on certain systems, see #12405.)
Until they can be fixed, they are being backed out.
- When in non-interactive mode, i.e. saving a crash report, don't
allow the image debug info loader to automatically grab missing but
available info packages. Otherwise we potentially download very large
packages with the user being entirely unaware, i.e. the 200MB debug
information package now present by default for gcc7's libgcc. This
should eventually be made a configurable preference though.
* define compat_thread_info, compat_rlim_t, compat_rlimit and
compat_thread_creation_attributes to be used when applicable in compatibility
mode.
* handle 32-bit types in _user_spawn_thread(), _user_get_thread_info(),
_user_get_next_thread_info(), _user_getrlimit(), _user_setrlimit(),
other syscalls are compatible as is.
* init TLS for compatibility mode threads.
Change-Id: I483ba95e6198ddac9d240671bcb56fcd2ad831d2
* in load_image_internal(), elf32_load_user_image checks whether the binary
format requires the compatibility mode.
* we then set up the flag THREAD_FLAGS_COMPAT_MODE and the address space size.
* the compatibility mode runtime_loader is hardcoded with x86/runtime_loader.
* if needed, the 64-bit flat_args structure is converted in-place to its 32-bit
layout.
* a 32-bit flat_args isn't handled yet (a 32-bit team execs a 64-bit binary).
Change-Id: Ia6a066bde8d1774d85de29b48dc500e27ae9668f
* define compat_area_info to be used when applicable in
compatibility mode.
* handle 32-bit types in _user_reserve_address_range(), _user_get_area_info(),
_user_get_next_area_info(), _user_transfer_area(), _user_clone_area(),
_user_create_area(), _user_map_file(), other syscalls are compatible as is.
* _get_next_area_info() doesn't work well with a 32-bit address cookie (address
could be in 64-bit range). Instead use _compat_get_next_area_info() which uses
the area id as cookie, though the areas are not ordered by address any more.
Change-Id: Ic7519ca8824aa2d534b0f03ea75a1bf6ae321535
* handle 32-bit types in _user_send_signal(), _user_sigaction(), _user_sigwait(),
_user_set_signal_stack(), _user_restore_signal_frame(), other syscalls are
compatible as is.
Change-Id: I4c8dc47bfa80f36e363d444d2a5a7be6c621606d
* define compat_image_info, compat_extended_image_info
to be used for respective 32-bit types of syscalls in compatibility mode.
* handle 32-bit types in _user_register_image, _user_get_image_info,
_user_get_next_image_info, other syscalls are compatible as is.
Change-Id: Ibbd33e6796208dfa70d869e36bf745bc3e18d330
* define compat_flock, compat_timespec, compat_stat, compat_attr_info,
compat_fs_info, compat_fd_info to be used for respective 32-bit types
of syscalls in compatibility mode.
* handle 32-bit types in common_fcntl(), _user_read_stat(), _user_stat_attr(),
_user_read_index_stat, _user_read_fs_info, _user_write_fs_info,
_user_get_next_fd_info, other syscalls are compatible as is.
Change-Id: I5b372169fe142f67b81fd6c27e0627d5119ba687
Completely unmodified and not wired into anything. Since my Haiku-specific
changes will go through Gerrit, it makes sense to import this first, so that
the diffs are readable.
* With the previous commit, we can now move functions that require
calling mmu_map_physical_memory to where they should have been
originally. This also allows the SMP safe mode menu entries to
be properly generated, now that smp_init is called prior to
main().
Change-Id: I05ddca5273b11cb4846021664c1ea2cf8ba723b7
* mmu_map_physical_address will get called prior to calling main()
which leaves us without a heap, malloc, and new. Instead, use
the kernel args physical allocated range array, and then
convert to our allocated memory region type on-demand.
Change-Id: I265fd165ef7143681e8e40c3686fda1a583c20dc
