lwn.git/arch/parisc/math-emu, branch docs-fixes

parisc: Fix double SIGFPE crash

2025-05-04T15:30:03+00:00

Camm noticed that on parisc a SIGFPE exception will crash an application with a second SIGFPE in the signal handler. Dave analyzed it, and it happens because glibc uses a double-word floating-point store to atomically update function descriptors. As a result of lazy binding, we hit a floating-point store in fpe_func almost immediately. When the T bit is set, an assist exception trap occurs when when the co-processor encounters *any* floating-point instruction except for a double store of register %fr0. The latter cancels all pending traps. Let's fix this by clearing the Trap (T) bit in the FP status register before returning to the signal handler in userspace. The issue can be reproduced with this test program: root@parisc:~# cat fpe.c static void fpe_func(int sig, siginfo_t *i, void *v) { sigset_t set; sigemptyset(&set); sigaddset(&set, SIGFPE); sigprocmask(SIG_UNBLOCK, &set, NULL); printf("GOT signal %d with si_code %ld\n", sig, i->si_code); } int main() { struct sigaction action = { .sa_sigaction = fpe_func, .sa_flags = SA_RESTART|SA_SIGINFO }; sigaction(SIGFPE, &action, 0); feenableexcept(FE_OVERFLOW); return printf("%lf\n",1.7976931348623158E308*1.7976931348623158E308); } root@parisc:~# gcc fpe.c -lm root@parisc:~# ./a.out Floating point exception root@parisc:~# strace -f ./a.out execve("./a.out", ["./a.out"], 0xf9ac7034 /* 20 vars */) = 0 getrlimit(RLIMIT_STACK, {rlim_cur=8192*1024, rlim_max=RLIM_INFINITY}) = 0 ... rt_sigaction(SIGFPE, {sa_handler=0x1110a, sa_mask=[], sa_flags=SA_RESTART|SA_SIGINFO}, NULL, 8) = 0 --- SIGFPE {si_signo=SIGFPE, si_code=FPE_FLTOVF, si_addr=0x1078f} --- --- SIGFPE {si_signo=SIGFPE, si_code=FPE_FLTOVF, si_addr=0xf8f21237} --- +++ killed by SIGFPE +++ Floating point exception Signed-off-by: Helge Deller Suggested-by: John David Anglin Reported-by: Camm Maguire Cc: stable@vger.kernel.org

parisc/math-emu: Remove unused struct 'exc_reg'

2024-05-10T13:09:15+00:00

This has been here since pre-git. Build tested. Signed-off-by: Dr. David Alan Gilbert Signed-off-by: Helge Deller

parisc: avoid c23 'nullptr' idenitifier

2024-02-27T21:51:44+00:00

Starting in c23, this is a reserved keyword, so in the future, using it will start causing build failures: arch/parisc/math-emu/frnd.c:36:23: error: expected ';', ',' or ')' before 'nullptr' Since I can't think of a good replacement name, add a leading underscore to the function argument to avoid this namespace conflict. Apparently all of these arguments are unused. Signed-off-by: Arnd Bergmann Signed-off-by: Helge Deller

parisc: math-emu: Avoid compiler warnings with W=1

2023-07-03T16:56:03+00:00

The math-emu code is a snapshot from the HP-UX kernel. They've been modified as little as possible. See arch/parisc/math-emu/README. Signed-off-by: Helge Deller

parisc: align '*' in comment in math-emu code

2022-06-25T22:19:27+00:00

Signed-off-by: Jiang Jian Reviewed-by: Kees Cook Signed-off-by: Helge Deller

parisc: Fix typos in comments

2022-05-08T18:01:12+00:00

Various spelling mistakes in comments. Detected with the help of Coccinelle. Signed-off-by: Julia Lawall Signed-off-by: Helge Deller

parisc: math-emu: Fix fall-through warnings

2021-09-01T20:18:18+00:00

Fix lots of fallthrough warnings, e.g.: arch/parisc/math-emu/fpudispatch.c:323:33: warning: this statement may fall through [-Wimplicit-fallthrough=] Signed-off-by: Helge Deller

parisc: math-emu: Avoid "fmt" macro collision

2021-08-30T08:18:24+00:00

The printk "fmt" macro was colliding. Rename like the others with a "bits" suffix. Fixes a build failure: arch/parisc/math-emu/decode_exc.c: In function 'decode_fpu': arch/parisc/math-emu/decode_exc.c:49:14: error: expected identifier before numeric constant 49 | #define fmt 11 /* bits 19 & 20 */ | ^~ ./include/linux/printk.h:379:6: note: in expansion of macro 'fmt' 379 | .fmt = __builtin_constant_p(_fmt) ? (_fmt) : NULL, \ | ^~~ ./include/linux/printk.h:417:3: note: in expansion of macro '__printk_index_emit' 417 | __printk_index_emit(_fmt, NULL, NULL); \ | ^~~~~~~~~~~~~~~~~~~ ./include/linux/printk.h:446:26: note: in expansion of macro 'printk_index_wrap' 446 | #define printk(fmt, ...) printk_index_wrap(_printk, fmt, ##__VA_ARGS__) | ^~~~~~~~~~~~~~~~~ arch/parisc/math-emu/decode_exc.c:339:3: note: in expansion of macro 'printk' 339 | printk("%s(%d) Unknown FPU exception 0x%x\n", __FILE__, | ^~~~~~ Cc: "James E.J. Bottomley" Cc: Helge Deller Cc: linux-parisc@vger.kernel.org Signed-off-by: Kees Cook Signed-off-by: Helge Deller

parisc: math-emu: Few spelling fixes in the file fpu.h

2021-04-06T13:34:14+00:00

with some additional cleanups by Helge. Signed-off-by: Bhaskar Chowdhury Acked-by: Randy Dunlap Signed-off-by: Helge Deller

Makefile: Convert -Wimplicit-fallthrough=3 to just -Wimplicit-fallthrough for clang

2019-08-10T19:45:37+00:00

A compilation -Wimplicit-fallthrough warning was enabled by commit a035d552a93b ("Makefile: Globally enable fall-through warning") Even though clang 10.0.0 does not currently support this warning without a patch, clang currently does not support a value for this option. Link: https://bugs.llvm.org/show_bug.cgi?id=39382 The gcc default for this warning is 3 so removing the =3 has no effect for gcc and enables the warning for patched versions of clang. Also remove the =3 from an existing use in a parisc Makefile: arch/parisc/math-emu/Makefile Signed-off-by: Joe Perches Reviewed-and-tested-by: Nathan Chancellor Cc: Gustavo A. R. Silva Signed-off-by: Linus Torvalds