Loongarch dev#54
Open
zhangyanyue01 wants to merge 128 commits into
Open
Conversation
zhangyanyue01
commented
Oct 10, 2024
zhangyanyue01
commented
- Add Loongarch support for trinity.
- Fix some errors which may cause SEGV.
- This code has been tested in 6.6 version kernel.
Still needs some cleanup, but this is a braindump of my private TODO that has a mix of old/new stuff.
These are usually only seen by libc, and never passed on to apps, but Trinity is doing raw syscalls, so we have to worry about these.
This patch brings perf_event_open() support up to date as of 4.14-rc6. Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>
In file included from ioctls/autofs.c:4:0:
ioctls/autofs.c:241:8: error: 'AUTOFS_IOC_EXPIRE_INDIRECT' undeclared here (not in a function); did you mean 'AUTOFS_IOC_EXPIRE_CMD'?
IOCTL(AUTOFS_IOC_EXPIRE_INDIRECT),
^
include/ioctls.h:53:15: note: in definition of macro 'IOCTL'
{ .request = _request, .name = #_request, }
^~~~~~~~
ioctls/autofs.c:242:8: error: 'AUTOFS_IOC_EXPIRE_DIRECT' undeclared here (not in a function); did you mean 'AUTOFS_IOC_EXPIRE_INDIRECT'?
IOCTL(AUTOFS_IOC_EXPIRE_DIRECT),
^
include/ioctls.h:53:15: note: in definition of macro 'IOCTL'
{ .request = _request, .name = #_request, }
^~~~~~~~
Define them as AUTOFS_IOC_EXPIRE_MULTI as they used to be.
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Commit 1f28c5d055032e7e8ee5e48198dca7e125d0eec6
("autofs: remove unused AUTOFS_IOC_EXPIRE_DIRECT/INDIRECT")
in the linux kernel removed the 2 defines
AUTOFS_IOC_EXPIRE_INDIRECT
AUTOFS_IOC_EXPIRE_DIRECT
Since they are used unconditionally the build fails.
To fix this add #ifdef/#endif around it.
Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Update and add some comments on the system call tables for s390x (64bit) and s390 (31bit). Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Add support for s390x (64 bit) and s390 (31 bit) support for system calls preadv2 and pwritev2. Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Add support for s390x (64 bit) and s390 (31 bit) support for system call statx. Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Add support for s390x (64 bit) and s390 (31 bit) support for system call mlock2. Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Add support for s390x (64 bit) and s390 (31 bit) support for system call copy_file_range. Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Add support for system call readdir on s390 (31 bit). Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
This patch adds support for the s390 specific system call runtime_instr() for 31 and 64 bit. Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Add support for s390 specific system calls s390_pci_mmio_read and 390_pci_mmio_write for 31 bit and 64 bit. Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Add support for s390 specific system call s390_guarded_storage for 31 bit and 64 bit. Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Add support for s390 specific system call s390_sthyi for 31 bit and 64 bit. Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
When a child process tests timer_create interface with sigev->sigev_signo equal to SIGINT, and then timer_settime let the timer go.When the timer has expired, a SIGINT would send to our child process. But SIGINT signal means ctrl-c is pressed,and then trinity-main will exit. To Avoid it, do not let timer_create create a timer with sigev->sigev_signo equal to SIGINT. Change-Id: I3aa1b99ff447a4a0a8bc2a9d33cf91f7e48cd275 Signed-off-by: Hongchen Zhang <zhanghongchen@loongson.cn>
This only adds support for riscv64, riscv32 is not included. The riscv64 syscall table is very close to aarch64 with small difference. Signed-off-by: Changbin Du <changbin.du@gmail.com>
fds/drm.c: In function ‘drmfd_dump’:
fds/drm.c:31:49: warning: unused parameter ‘global’ [-Wunused-parameter]
31 | static void drmfd_dump(struct object *obj, bool global)
| ~~~~~^~~~~~
Print 'global' to fix warning.
Trinity on cpus which have many cores may open large number of child->pidstatfile, and may be reach the limition(1024 on most platforms), so just panic and warn user to increase the limition.
nukes those warnings like.. main.c:467:2: warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement] because it isn't 1990 any more.
this needs >10000 threads, but gcc has started warning, so this shuts it up
Both sockaddr_pn and sockaddr_pppox are defined as "packed"
on x86-64, unlike struct sockaddr, which generates warnings
like this:
net/proto-phonet.c: In function ‘phonet_gen_sockaddr’:
net/proto-phonet.c:22:18: warning: converting a packed ‘struct sockaddr_pn’ pointer (alignment 1) to a ‘struct sockaddr’ pointer (alignment 2) may result in an unaligned pointer value [-Waddress-of-packed-member]
22 | *addr = (struct sockaddr *) pn;
| ^~~~~~~~
In file included from net/proto-phonet.c:5:
/usr/include/linux/phonet.h:102:8: note: defined here
102 | struct sockaddr_pn {
| ^~~~~~~~~~~
In file included from /usr/include/x86_64-linux-gnu/sys/socket.h:33,
from net/proto-phonet.c:2:
/usr/include/x86_64-linux-gnu/bits/socket.h:178:8: note: defined here
178 | struct sockaddr
| ^~~~~~~~
I'm pretty sure the alignment is fine as other things would
be broken if they weren't, so just ignore the warnings in
these two files.
Drop decnet as it has been removed since kernel 6.1 and torvalds/linux@1202cdd resulting in the following build failure: net/proto-decnet.c:5:10: fatal error: linux/dn.h: No such file or directory 5 | #include <linux/dn.h> | ^~~~~~~~~~~~ Fixes: - http://autobuild.buildroot.org/results/47e0a5e0b6fcf33ab4f9848d5d8c2be9e5283950 Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
…te. I met this under mips64 debian
It was dropped in upstream in commit 61e15f871241 (KVM: Delete all references to removed KVM_SET_MEMORY_REGION ioctl). Co-authored-by: Jiri Slaby <jslaby@suse.cz>
…viders. (kernelslacker#43) Signed-off-by: Weisson <hanxiaobupt@bupt.cn> Co-authored-by: Weisson <wenshu.hx@alibaba-inc.com>
kernel v6.3-rc1 removed some drm drivers, so add the extra check to make the compile succeed. 7872bc2cb13e drm: Remove the obsolete driver-savage 28483b8666bf drm: Remove the obsolete driver-r128 96ed7db55bef drm: Remove the obsolete driver-mga cab18866fead drm: Remove the obsolete driver-i810 Signed-off-by: Chunyu Hu <chuhu@redhat.com>
Some IOCTL_NOUVEAU tests are checked under USE_DRM_MGA macro, this is not correct, correct it. Fixes: 6823bac (ioctls/drm: check i810,mga,r128 and savage support (kernelslacker#44)) Signed-off-by: Chunyu Hu <chuhu@redhat.com>
* compat: add deprecated KVM IOCTL definitions They were dropped in 6.8 by commit a5d3df8ae13f (KVM: remove deprecated UAPIs). * compat: add removed KVM structures They were dropped in 6.9 in c0a411904e15 (KVM: remove more traces of device assignment UAPI). * ioctls/kvm: make s390 ioctls s390-only They are now s390-only since commit 71cd774ad2f9 (KVM: s390: move s390-specific structs to uapi/asm/kvm.h). * ioctls/kvm: make powerpc ioctls powerpc-only They are now powerpc-only since commit d750951c9ed7 (KVM: powerpc: move powerpc-specific structs to uapi/asm/kvm.h). --------- Co-authored-by: Jiri Slaby <jslaby@suse.cz>
from man page, it requires _GNU_SOURCE, and it requires two parameters,
so it would not be able to pass the configure check. It works with older gcc
as old gcc allows conflict on 'static' keyword.
But it would hit copmile broken if not fixed on fedora 40. Because gcc
version would check the declaration conflict on static function. In header
it's not static, but in trinity, it's declared as static.
The 'ifndef memfd_create' don't work because memfd_create is not a macro,
so remove it.
Before the fix:
[trinity.git]$ cat /etc/redhat-release
Fedora release 40 (Forty)
[trinity.git]$ ./configure | grep memfd_create
[*] Checking if glibc provides memfd_create.. [NO]
[chuhu@dell-per7425-02 trinity.git]$ make -j8
CC arg-decoder.o
CC blockdevs.o
CC child.o
CC debug.o
CC devices.o
CC ftrace.o
CC generate-args.o
CC kcov.o
CC locks.o
CC log-files.o
CC log.o
CC main.o
CC objects.o
CC output.o
CC params.o
CC pathnames.o
CC pids.o
CC post-mortem.o
CC random-syscall.o
CC results.o
CC shm.o
CC signals.o
CC stats.o
CC syscall.o
CC sysv-shm.o
CC tables-biarch.o
CC tables-uniarch.o
CC tables.o
CC taint.o
CC trinity.o
CC uid.o
CC utils.o
CC fds/bpf.o
CC fds/drm.o
CC fds/epoll.o
CC fds/eventfd.o
CC fds/fanotify_init.o
CC fds/fds.o
CC fds/files.o
CC fds/inotify.o
CC fds/memfd.o
CC fds/perf.o
CC fds/pipes.o
CC fds/sockets.o
CC fds/testfiles.o
fds/memfd.c:22:12: error: static declaration of ‘memfd_create’ follows non-static declaration
22 | static int memfd_create(__unused__ const char *uname, __unused__ unsigned int flag)
| ^~~~~~~~~~~~
In file included from /usr/include/bits/mman-linux.h:116,
from /usr/include/bits/mman.h:38,
from /usr/include/sys/mman.h:41,
from fds/memfd.c:8:
/usr/include/bits/mman-shared.h:55:5: note: previous declaration of ‘memfd_create’ with type ‘int(const char *, unsigned int)’
55 | int memfd_create (const char *__name, unsigned int __flags) __THROW;
| ^~~~~~~~~~~~
make: *** [Makefile:113: fds/memfd.o] Error 1
make: *** Waiting for unfinished jobs....
After the fix:
[trinity]$ ./configure | grep memfd_create
[*] Checking if glibc provides memfd_create.. [YES]
[trinity]$ make -j8 | grep memfd_create
CC syscalls/memfd_create.o
Signed-off-by: Chunyu Hu <chuhu@redhat.com>
I'm never realistically going to spend a huge amount of time on this tool again, so remove this list of stuff that isn't going to happen.
This was never implemented, and is aparently broken on 32bit (kernelslacker#52)
since 460be1d527a8 ("drm/nouveau: move more missing UAPI bits") in
v6.10-rc1, the two structs are moved to the uapi in
include/drm/drm_nouveau.h (drm_nouveau_notifierobj_alloc and
drm_nouveau_gpuobj_free), so the definition of the macros and the
structs are not in the same place any more, so can't use the macros
to determine if the two structs are defined. So change to compile in
configure to determine that.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
16 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.