* [PATCH v2 02/23] linux-user/aarch64: Implement setup_sigtramp [not found] <20210618192951.125651-1-richard.henderson@linaro.org> @ 2021-06-18 19:29 ` Richard Henderson 2021-06-29 13:36 ` Peter Maydell 2021-06-18 19:29 ` [PATCH v2 03/23] linux-user/arm: Split out v2_frame Richard Henderson ` (2 subsequent siblings) 3 siblings, 1 reply; 11+ messages in thread From: Richard Henderson @ 2021-06-18 19:29 UTC (permalink / raw) To: qemu-devel; +Cc: laurent, alex.bennee, qemu-arm, Philippe Mathieu-Daudé Create and record the rt signal trampoline. Use it when the guest does not use SA_RESTORER. Cc: qemu-arm@nongnu.org Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- linux-user/aarch64/target_signal.h | 2 ++ linux-user/aarch64/signal.c | 28 ++++++++++++++++++---------- 2 files changed, 20 insertions(+), 10 deletions(-) diff --git a/linux-user/aarch64/target_signal.h b/linux-user/aarch64/target_signal.h index 18013e1b23..7580d99403 100644 --- a/linux-user/aarch64/target_signal.h +++ b/linux-user/aarch64/target_signal.h @@ -25,4 +25,6 @@ typedef struct target_sigaltstack { #define TARGET_SEGV_MTESERR 9 /* Synchronous ARM MTE exception */ #define TARGET_ARCH_HAS_SETUP_FRAME +#define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 + #endif /* AARCH64_TARGET_SIGNAL_H */ diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c index 662bcd1c4e..65b84eb04e 100644 --- a/linux-user/aarch64/signal.c +++ b/linux-user/aarch64/signal.c @@ -108,7 +108,6 @@ struct target_rt_sigframe { struct target_rt_frame_record { uint64_t fp; uint64_t lr; - uint32_t tramp[2]; }; static void target_setup_general_frame(struct target_rt_sigframe *sf, @@ -495,15 +494,7 @@ static void target_setup_frame(int usig, struct target_sigaction *ka, if (ka->sa_flags & TARGET_SA_RESTORER) { return_addr = ka->sa_restorer; } else { - /* - * mov x8,#__NR_rt_sigreturn; svc #0 - * Since these are instructions they need to be put as little-endian - * regardless of target default or current CPU endianness. - */ - __put_user_e(0xd2801168, &fr->tramp[0], le); - __put_user_e(0xd4000001, &fr->tramp[1], le); - return_addr = frame_addr + fr_ofs - + offsetof(struct target_rt_frame_record, tramp); + return_addr = default_rt_sigreturn; } env->xregs[0] = usig; env->xregs[29] = frame_addr + fr_ofs; @@ -576,3 +567,20 @@ long do_sigreturn(CPUARMState *env) { return do_rt_sigreturn(env); } + +void setup_sigtramp(abi_ulong sigtramp_page) +{ + uint32_t *tramp = lock_user(VERIFY_WRITE, sigtramp_page, 8, 0); + assert(tramp != NULL); + + /* + * mov x8,#__NR_rt_sigreturn; svc #0 + * Since these are instructions they need to be put as little-endian + * regardless of target default or current CPU endianness. + */ + __put_user_e(0xd2801168, &tramp[0], le); + __put_user_e(0xd4000001, &tramp[1], le); + + default_rt_sigreturn = sigtramp_page; + unlock_user(tramp, sigtramp_page, 8); +} -- 2.25.1 ^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [PATCH v2 02/23] linux-user/aarch64: Implement setup_sigtramp 2021-06-18 19:29 ` [PATCH v2 02/23] linux-user/aarch64: Implement setup_sigtramp Richard Henderson @ 2021-06-29 13:36 ` Peter Maydell 2021-07-01 19:27 ` Richard Henderson 0 siblings, 1 reply; 11+ messages in thread From: Peter Maydell @ 2021-06-29 13:36 UTC (permalink / raw) To: Richard Henderson Cc: Philippe Mathieu-Daudé, qemu-arm, QEMU Developers, Laurent Vivier On Fri, 18 Jun 2021 at 20:33, Richard Henderson <richard.henderson@linaro.org> wrote: > > Create and record the rt signal trampoline. > Use it when the guest does not use SA_RESTORER. > > Cc: qemu-arm@nongnu.org > Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org> > Signed-off-by: Richard Henderson <richard.henderson@linaro.org> > --- > linux-user/aarch64/target_signal.h | 2 ++ > linux-user/aarch64/signal.c | 28 ++++++++++++++++++---------- > 2 files changed, 20 insertions(+), 10 deletions(-) > > diff --git a/linux-user/aarch64/target_signal.h b/linux-user/aarch64/target_signal.h > index 18013e1b23..7580d99403 100644 > --- a/linux-user/aarch64/target_signal.h > +++ b/linux-user/aarch64/target_signal.h > @@ -25,4 +25,6 @@ typedef struct target_sigaltstack { > #define TARGET_SEGV_MTESERR 9 /* Synchronous ARM MTE exception */ > > #define TARGET_ARCH_HAS_SETUP_FRAME > +#define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 > + > #endif /* AARCH64_TARGET_SIGNAL_H */ > diff --git a/linux-user/aarch64/signal.c b/linux-user/aarch64/signal.c > index 662bcd1c4e..65b84eb04e 100644 > --- a/linux-user/aarch64/signal.c > +++ b/linux-user/aarch64/signal.c > @@ -108,7 +108,6 @@ struct target_rt_sigframe { > struct target_rt_frame_record { > uint64_t fp; > uint64_t lr; > - uint32_t tramp[2]; > }; > > static void target_setup_general_frame(struct target_rt_sigframe *sf, > @@ -495,15 +494,7 @@ static void target_setup_frame(int usig, struct target_sigaction *ka, A little way up from here there's a comment: /* Reserve space for the return code. On a real system this would * be within the VDSO. So, despite the name this is not a "real" * record within the frame. */ fr_ofs = layout.total_size; layout.total_size += sizeof(struct target_rt_frame_record); That now needs updating, because we're no longer putting the return code in that target_rt_frame_record. I think the 'struct target_rt_frame_record' now does correspond to the kernel's 'struct frame_record', right? Otherwise Reviewed-by: Peter Maydell <peter.maydell@linaro.org> thanks -- PMM ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH v2 02/23] linux-user/aarch64: Implement setup_sigtramp 2021-06-29 13:36 ` Peter Maydell @ 2021-07-01 19:27 ` Richard Henderson 0 siblings, 0 replies; 11+ messages in thread From: Richard Henderson @ 2021-07-01 19:27 UTC (permalink / raw) To: Peter Maydell Cc: Philippe Mathieu-Daudé, qemu-arm, QEMU Developers, Laurent Vivier On 6/29/21 6:36 AM, Peter Maydell wrote: > A little way up from here there's a comment: > > /* Reserve space for the return code. On a real system this would > * be within the VDSO. So, despite the name this is not a "real" > * record within the frame. > */ > fr_ofs = layout.total_size; > layout.total_size += sizeof(struct target_rt_frame_record); > > That now needs updating, because we're no longer putting the return > code in that target_rt_frame_record. Updated to /* * Reserve space for the standard frame unwind pair: fp, lr. * Despite the name this is not a "real" record within the frame. */ > I think the 'struct target_rt_frame_record' now does correspond > to the kernel's 'struct frame_record', right? Yes. r~ ^ permalink raw reply [flat|nested] 11+ messages in thread
* [PATCH v2 03/23] linux-user/arm: Split out v2_frame [not found] <20210618192951.125651-1-richard.henderson@linaro.org> 2021-06-18 19:29 ` [PATCH v2 02/23] linux-user/aarch64: Implement setup_sigtramp Richard Henderson @ 2021-06-18 19:29 ` Richard Henderson 2021-06-29 13:53 ` Peter Maydell 2021-06-18 19:29 ` [PATCH v2 04/23] linux-user/arm: Force v2 frames for fdpic Richard Henderson 2021-06-18 19:29 ` [PATCH v2 05/23] linux-user/arm: Implement setup_sigtramp Richard Henderson 3 siblings, 1 reply; 11+ messages in thread From: Richard Henderson @ 2021-06-18 19:29 UTC (permalink / raw) To: qemu-devel; +Cc: laurent, alex.bennee, qemu-arm Split out a helper function to test for a v2 signal frame. Cc: qemu-arm@nongnu.org Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- linux-user/arm/signal.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/linux-user/arm/signal.c b/linux-user/arm/signal.c index 32b68ee302..cb65623965 100644 --- a/linux-user/arm/signal.c +++ b/linux-user/arm/signal.c @@ -165,6 +165,11 @@ static inline int valid_user_regs(CPUARMState *regs) return 1; } +static bool v2_frame(void) +{ + return get_osversion() >= 0x020612; +} + static void setup_sigcontext(struct target_sigcontext *sc, /*struct _fpstate *fpstate,*/ CPUARMState *env, abi_ulong mask) @@ -422,7 +427,7 @@ sigsegv: void setup_frame(int usig, struct target_sigaction *ka, target_sigset_t *set, CPUARMState *regs) { - if (get_osversion() >= 0x020612) { + if (v2_frame()) { setup_frame_v2(usig, ka, set, regs); } else { setup_frame_v1(usig, ka, set, regs); @@ -516,7 +521,7 @@ void setup_rt_frame(int usig, struct target_sigaction *ka, target_siginfo_t *info, target_sigset_t *set, CPUARMState *env) { - if (get_osversion() >= 0x020612) { + if (v2_frame()) { setup_rt_frame_v2(usig, ka, info, set, env); } else { setup_rt_frame_v1(usig, ka, info, set, env); @@ -734,7 +739,7 @@ badframe: long do_sigreturn(CPUARMState *env) { - if (get_osversion() >= 0x020612) { + if (v2_frame()) { return do_sigreturn_v2(env); } else { return do_sigreturn_v1(env); @@ -823,7 +828,7 @@ badframe: long do_rt_sigreturn(CPUARMState *env) { - if (get_osversion() >= 0x020612) { + if (v2_frame()) { return do_rt_sigreturn_v2(env); } else { return do_rt_sigreturn_v1(env); -- 2.25.1 ^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [PATCH v2 03/23] linux-user/arm: Split out v2_frame 2021-06-18 19:29 ` [PATCH v2 03/23] linux-user/arm: Split out v2_frame Richard Henderson @ 2021-06-29 13:53 ` Peter Maydell 2021-06-29 14:30 ` Richard Henderson 0 siblings, 1 reply; 11+ messages in thread From: Peter Maydell @ 2021-06-29 13:53 UTC (permalink / raw) To: Richard Henderson Cc: QEMU Developers, qemu-arm, Alex Bennée, Laurent Vivier On Fri, 18 Jun 2021 at 20:32, Richard Henderson <richard.henderson@linaro.org> wrote: > > Split out a helper function to test for a v2 signal frame. > > Cc: qemu-arm@nongnu.org > Signed-off-by: Richard Henderson <richard.henderson@linaro.org> > --- > linux-user/arm/signal.c | 13 +++++++++---- > 1 file changed, 9 insertions(+), 4 deletions(-) > > diff --git a/linux-user/arm/signal.c b/linux-user/arm/signal.c > index 32b68ee302..cb65623965 100644 > --- a/linux-user/arm/signal.c > +++ b/linux-user/arm/signal.c > @@ -165,6 +165,11 @@ static inline int valid_user_regs(CPUARMState *regs) > return 1; > } > > +static bool v2_frame(void) > +{ > + return get_osversion() >= 0x020612; > +} Not sure how much we care about supporting claiming to be a 15-year-old kernel any more (especially since we set UNAME_MINIMUM_RELEASE to 2.6.32...) so it's awfully tempting to just blow away the v1 frame support instead... Anyway Reviewed-by: Peter Maydell <peter.maydell@linaro.org> thanks -- PMM ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH v2 03/23] linux-user/arm: Split out v2_frame 2021-06-29 13:53 ` Peter Maydell @ 2021-06-29 14:30 ` Richard Henderson 0 siblings, 0 replies; 11+ messages in thread From: Richard Henderson @ 2021-06-29 14:30 UTC (permalink / raw) To: Peter Maydell; +Cc: QEMU Developers, qemu-arm, Alex Bennée, Laurent Vivier On 6/29/21 6:53 AM, Peter Maydell wrote: > On Fri, 18 Jun 2021 at 20:32, Richard Henderson > <richard.henderson@linaro.org> wrote: >> >> Split out a helper function to test for a v2 signal frame. >> >> Cc: qemu-arm@nongnu.org >> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> >> --- >> linux-user/arm/signal.c | 13 +++++++++---- >> 1 file changed, 9 insertions(+), 4 deletions(-) >> >> diff --git a/linux-user/arm/signal.c b/linux-user/arm/signal.c >> index 32b68ee302..cb65623965 100644 >> --- a/linux-user/arm/signal.c >> +++ b/linux-user/arm/signal.c >> @@ -165,6 +165,11 @@ static inline int valid_user_regs(CPUARMState *regs) >> return 1; >> } >> >> +static bool v2_frame(void) >> +{ >> + return get_osversion() >= 0x020612; >> +} > > Not sure how much we care about supporting claiming to be a 15-year-old > kernel any more (especially since we set UNAME_MINIMUM_RELEASE to 2.6.32...) > so it's awfully tempting to just blow away the v1 frame support instead... > > Anyway > Reviewed-by: Peter Maydell <peter.maydell@linaro.org> I hadn't noticed the minimum release setting. Yes, I think it would be better to remove the v1 support instead of bodging around it like this. r~ ^ permalink raw reply [flat|nested] 11+ messages in thread
* [PATCH v2 04/23] linux-user/arm: Force v2 frames for fdpic [not found] <20210618192951.125651-1-richard.henderson@linaro.org> 2021-06-18 19:29 ` [PATCH v2 02/23] linux-user/aarch64: Implement setup_sigtramp Richard Henderson 2021-06-18 19:29 ` [PATCH v2 03/23] linux-user/arm: Split out v2_frame Richard Henderson @ 2021-06-18 19:29 ` Richard Henderson 2021-06-29 13:54 ` Peter Maydell 2021-06-18 19:29 ` [PATCH v2 05/23] linux-user/arm: Implement setup_sigtramp Richard Henderson 3 siblings, 1 reply; 11+ messages in thread From: Richard Henderson @ 2021-06-18 19:29 UTC (permalink / raw) To: qemu-devel; +Cc: laurent, alex.bennee, qemu-arm The value of get_os_release may be controlled by a command line option. Since fdpic was added in v4.14, and v2 frame were added in v2.6.12, this makes no change under normal conditions. Cc: qemu-arm@nongnu.org Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- linux-user/arm/signal.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/linux-user/arm/signal.c b/linux-user/arm/signal.c index cb65623965..2d30345fc2 100644 --- a/linux-user/arm/signal.c +++ b/linux-user/arm/signal.c @@ -167,7 +167,14 @@ static inline int valid_user_regs(CPUARMState *regs) static bool v2_frame(void) { - return get_osversion() >= 0x020612; + /* + * We do not create fdpic trampolines for v1 frames. + * Thus we force v2 frames, regardless of what uname says. + * Support for fdpic dates from Linux 4.14, so this is not + * really a behaviour change. + */ + int is_fdpic = info_is_fdpic(((TaskState *)thread_cpu->opaque)->info); + return is_fdpic || get_osversion() >= 0x020612; } static void -- 2.25.1 ^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [PATCH v2 04/23] linux-user/arm: Force v2 frames for fdpic 2021-06-18 19:29 ` [PATCH v2 04/23] linux-user/arm: Force v2 frames for fdpic Richard Henderson @ 2021-06-29 13:54 ` Peter Maydell 0 siblings, 0 replies; 11+ messages in thread From: Peter Maydell @ 2021-06-29 13:54 UTC (permalink / raw) To: Richard Henderson Cc: QEMU Developers, qemu-arm, Alex Bennée, Laurent Vivier On Fri, 18 Jun 2021 at 20:32, Richard Henderson <richard.henderson@linaro.org> wrote: > > The value of get_os_release may be controlled by a command > line option. Since fdpic was added in v4.14, and v2 frame > were added in v2.6.12, this makes no change under normal conditions. > > Cc: qemu-arm@nongnu.org > Signed-off-by: Richard Henderson <richard.henderson@linaro.org> Reviewed-by: Peter Maydell <peter.maydell@linaro.org> thanks -- PMM ^ permalink raw reply [flat|nested] 11+ messages in thread
* [PATCH v2 05/23] linux-user/arm: Implement setup_sigtramp [not found] <20210618192951.125651-1-richard.henderson@linaro.org> ` (2 preceding siblings ...) 2021-06-18 19:29 ` [PATCH v2 04/23] linux-user/arm: Force v2 frames for fdpic Richard Henderson @ 2021-06-18 19:29 ` Richard Henderson 2021-06-29 14:09 ` Peter Maydell 3 siblings, 1 reply; 11+ messages in thread From: Richard Henderson @ 2021-06-18 19:29 UTC (permalink / raw) To: qemu-devel; +Cc: laurent, alex.bennee, qemu-arm ARM is more complicated than the others, in that we also have trampolines for using SA_RESTORER with FDPIC, and we need to create trampolines for both ARM and Thumb modes. Cc: qemu-arm@nongnu.org Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- linux-user/arm/target_signal.h | 2 + linux-user/arm/signal.c | 170 +++++++++++++++++++-------------- 2 files changed, 100 insertions(+), 72 deletions(-) diff --git a/linux-user/arm/target_signal.h b/linux-user/arm/target_signal.h index 0998dd6dfa..1e7fb0cecb 100644 --- a/linux-user/arm/target_signal.h +++ b/linux-user/arm/target_signal.h @@ -22,4 +22,6 @@ typedef struct target_sigaltstack { #include "../generic/signal.h" #define TARGET_ARCH_HAS_SETUP_FRAME +#define TARGET_ARCH_HAS_SIGTRAMP_PAGE 1 + #endif /* ARM_TARGET_SIGNAL_H */ diff --git a/linux-user/arm/signal.c b/linux-user/arm/signal.c index 2d30345fc2..63bdd59ab9 100644 --- a/linux-user/arm/signal.c +++ b/linux-user/arm/signal.c @@ -101,13 +101,12 @@ struct sigframe_v1 { struct target_sigcontext sc; abi_ulong extramask[TARGET_NSIG_WORDS-1]; - abi_ulong retcode[4]; }; struct sigframe_v2 { struct target_ucontext_v2 uc; - abi_ulong retcode[4]; + abi_ulong fdpic_ret; }; struct rt_sigframe_v1 @@ -116,49 +115,20 @@ struct rt_sigframe_v1 abi_ulong puc; struct target_siginfo info; struct target_ucontext_v1 uc; - abi_ulong retcode[4]; }; struct rt_sigframe_v2 { struct target_siginfo info; struct target_ucontext_v2 uc; - abi_ulong retcode[4]; + abi_ulong fdpic_ret; }; /* - * For ARM syscalls, we encode the syscall number into the instruction. + * Stubs needed to make sure the FD register (r9) contains the right value. + * There are 4 of them, each consuming 8 bytes. */ -#define SWI_SYS_SIGRETURN (0xef000000|(TARGET_NR_sigreturn + ARM_SYSCALL_BASE)) -#define SWI_SYS_RT_SIGRETURN (0xef000000|(TARGET_NR_rt_sigreturn + ARM_SYSCALL_BASE)) - -/* - * For Thumb syscalls, we pass the syscall number via r7. We therefore - * need two 16-bit instructions. - */ -#define SWI_THUMB_SIGRETURN (0xdf00 << 16 | 0x2700 | (TARGET_NR_sigreturn)) -#define SWI_THUMB_RT_SIGRETURN (0xdf00 << 16 | 0x2700 | (TARGET_NR_rt_sigreturn)) - -static const abi_ulong retcodes[4] = { - SWI_SYS_SIGRETURN, SWI_THUMB_SIGRETURN, - SWI_SYS_RT_SIGRETURN, SWI_THUMB_RT_SIGRETURN -}; - -/* - * Stub needed to make sure the FD register (r9) contains the right - * value. - */ -static const unsigned long sigreturn_fdpic_codes[3] = { - 0xe59fc004, /* ldr r12, [pc, #4] to read function descriptor */ - 0xe59c9004, /* ldr r9, [r12, #4] to setup GOT */ - 0xe59cf000 /* ldr pc, [r12] to jump into restorer */ -}; - -static const unsigned long sigreturn_fdpic_thumb_codes[3] = { - 0xc008f8df, /* ldr r12, [pc, #8] to read function descriptor */ - 0x9004f8dc, /* ldr r9, [r12, #4] to setup GOT */ - 0xf000f8dc /* ldr pc, [r12] to jump into restorer */ -}; +static abi_ulong sigreturn_fdpic_tramp; static inline int valid_user_regs(CPUARMState *regs) { @@ -219,13 +189,12 @@ get_sigframe(struct target_sigaction *ka, CPUARMState *regs, int framesize) static int setup_return(CPUARMState *env, struct target_sigaction *ka, - abi_ulong *rc, abi_ulong frame_addr, int usig, abi_ulong rc_addr) + abi_ulong frame_addr, int usig) { abi_ulong handler = 0; abi_ulong handler_fdpic_GOT = 0; abi_ulong retcode; - - int thumb; + int thumb, retcode_idx; int is_fdpic = info_is_fdpic(((TaskState *)thread_cpu->opaque)->info); if (is_fdpic) { @@ -243,6 +212,7 @@ setup_return(CPUARMState *env, struct target_sigaction *ka, } thumb = handler & 1; + retcode_idx = thumb + (ka->sa_flags & TARGET_SA_SIGINFO ? 2 : 0); uint32_t cpsr = cpsr_read(env); @@ -260,37 +230,24 @@ setup_return(CPUARMState *env, struct target_sigaction *ka, if (ka->sa_flags & TARGET_SA_RESTORER) { if (is_fdpic) { - /* For FDPIC we ensure that the restorer is called with a - * correct r9 value. For that we need to write code on - * the stack that sets r9 and jumps back to restorer - * value. + /* + * For FDPIC we ensure that the restorer is called with a + * correct r9 value. For that we use a special trampoline + * that reads the function descriptor from the frame, + * sets r9 and jumps back to restorer value. */ - if (thumb) { - __put_user(sigreturn_fdpic_thumb_codes[0], rc); - __put_user(sigreturn_fdpic_thumb_codes[1], rc + 1); - __put_user(sigreturn_fdpic_thumb_codes[2], rc + 2); - __put_user((abi_ulong)ka->sa_restorer, rc + 3); - } else { - __put_user(sigreturn_fdpic_codes[0], rc); - __put_user(sigreturn_fdpic_codes[1], rc + 1); - __put_user(sigreturn_fdpic_codes[2], rc + 2); - __put_user((abi_ulong)ka->sa_restorer, rc + 3); - } - - retcode = rc_addr + thumb; + abi_ulong fd_ofs = (retcode_idx & 2 + ? offsetof(struct rt_sigframe_v2, fdpic_ret) + : offsetof(struct sigframe_v2, fdpic_ret)); + put_user_ual(ka->sa_restorer, frame_addr + fd_ofs); + /* Each trampoline variant consumes 8-byte slot. */ + retcode = sigreturn_fdpic_tramp + retcode_idx * 8 + thumb; } else { retcode = ka->sa_restorer; } } else { - unsigned int idx = thumb; - - if (ka->sa_flags & TARGET_SA_SIGINFO) { - idx += 2; - } - - __put_user(retcodes[idx], rc); - - retcode = rc_addr + thumb; + /* Each trampoline variant consumes one 4-byte slot. */ + retcode = default_sigreturn + retcode_idx * 4 + thumb; } env->regs[0] = usig; @@ -394,8 +351,7 @@ static void setup_frame_v1(int usig, struct target_sigaction *ka, __put_user(set->sig[i], &frame->extramask[i - 1]); } - if (setup_return(regs, ka, frame->retcode, frame_addr, usig, - frame_addr + offsetof(struct sigframe_v1, retcode))) { + if (setup_return(regs, ka, frame_addr, usig)) { goto sigsegv; } @@ -419,8 +375,7 @@ static void setup_frame_v2(int usig, struct target_sigaction *ka, setup_sigframe_v2(&frame->uc, set, regs); - if (setup_return(regs, ka, frame->retcode, frame_addr, usig, - frame_addr + offsetof(struct sigframe_v2, retcode))) { + if (setup_return(regs, ka, frame_addr, usig)) { goto sigsegv; } @@ -475,8 +430,7 @@ static void setup_rt_frame_v1(int usig, struct target_sigaction *ka, __put_user(set->sig[i], &frame->uc.tuc_sigmask.sig[i]); } - if (setup_return(env, ka, frame->retcode, frame_addr, usig, - frame_addr + offsetof(struct rt_sigframe_v1, retcode))) { + if (setup_return(env, ka, frame_addr, usig)) { goto sigsegv; } @@ -509,8 +463,7 @@ static void setup_rt_frame_v2(int usig, struct target_sigaction *ka, setup_sigframe_v2(&frame->uc, set, env); - if (setup_return(env, ka, frame->retcode, frame_addr, usig, - frame_addr + offsetof(struct rt_sigframe_v2, retcode))) { + if (setup_return(env, ka, frame_addr, usig)) { goto sigsegv; } @@ -841,3 +794,76 @@ long do_rt_sigreturn(CPUARMState *env) return do_rt_sigreturn_v1(env); } } + +void setup_sigtramp(abi_ulong sigtramp_page) +{ + enum { + /* For ARM, we encode the syscall number into the instruction. */ + SWI_SYS_SIGRETURN = + 0xef000000 | (TARGET_NR_sigreturn + ARM_SYSCALL_BASE), + SWI_SYS_RT_SIGRETURN = + 0xef000000 | (TARGET_NR_rt_sigreturn + ARM_SYSCALL_BASE), + + /* + * For Thumb , we pass the syscall number via r7. + * We therefore need two 16-bit instructions. + */ + SWI_THUMB_SIGRETURN = + 0xdf00 << 16 | 0x2700 | TARGET_NR_sigreturn, + SWI_THUMB_RT_SIGRETURN = + 0xdf00 << 16 | 0x2700 | TARGET_NR_rt_sigreturn, + + SIGFRAME_FDPIC_OFS = offsetof(struct sigframe_v2, fdpic_ret), + RT_SIGFRAME_FDPIC_OFS = offsetof(struct rt_sigframe_v2, fdpic_ret), + }; + + uint32_t total_size = 4 * 4 + 2 * 8; + uint32_t *tramp = lock_user(VERIFY_WRITE, sigtramp_page, total_size, 0); + uint32_t i = 0; + + assert(tramp != NULL); + + default_sigreturn = sigtramp_page; + __put_user(SWI_SYS_SIGRETURN, &tramp[i++]); + __put_user(SWI_THUMB_SIGRETURN, &tramp[i++]); + __put_user(SWI_SYS_RT_SIGRETURN, &tramp[i++]); + __put_user(SWI_THUMB_RT_SIGRETURN, &tramp[i++]); + + /* + * FDPIC require trampolines to call sa_restorer. + * + * ARM versions use: + * ldr r9, [sp, #ofs] + * ldmia r9, {r9, pc} + * + * Thumb versions use: + * ldrd r9, r10, [sp, #ofs] + * bx r10 + * nop + */ + sigreturn_fdpic_tramp = sigtramp_page + i * 4; + + /* ARM sigframe */ + QEMU_BUILD_BUG_ON(SIGFRAME_FDPIC_OFS > 0xfff); + __put_user(0xe59d9000 | SIGFRAME_FDPIC_OFS, &tramp[i++]); + __put_user(0xe8998200, &tramp[i++]); + + /* Thumb sigframe */ + QEMU_BUILD_BUG_ON(SIGFRAME_FDPIC_OFS > 0xff << 2); + QEMU_BUILD_BUG_ON(SIGFRAME_FDPIC_OFS & 3); + __put_user(0x9a00e9dd | (SIGFRAME_FDPIC_OFS << 14), &tramp[i++]); + __put_user(0x46c04750, &tramp[i++]); + + /* ARM rt_sigframe */ + QEMU_BUILD_BUG_ON(RT_SIGFRAME_FDPIC_OFS > 0xfff); + __put_user(0xe59d9000 | RT_SIGFRAME_FDPIC_OFS, &tramp[i++]); + __put_user(0xe8998200, &tramp[i++]); + + /* Thumb rt_sigframe */ + QEMU_BUILD_BUG_ON(RT_SIGFRAME_FDPIC_OFS > 0xff << 2); + QEMU_BUILD_BUG_ON(RT_SIGFRAME_FDPIC_OFS & 3); + __put_user(0x9a00e9dd | (RT_SIGFRAME_FDPIC_OFS << 14), &tramp[i++]); + __put_user(0x46c04750, &tramp[i++]); + + unlock_user(tramp, sigtramp_page, total_size); +} -- 2.25.1 ^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [PATCH v2 05/23] linux-user/arm: Implement setup_sigtramp 2021-06-18 19:29 ` [PATCH v2 05/23] linux-user/arm: Implement setup_sigtramp Richard Henderson @ 2021-06-29 14:09 ` Peter Maydell 2021-06-29 18:32 ` Richard Henderson 0 siblings, 1 reply; 11+ messages in thread From: Peter Maydell @ 2021-06-29 14:09 UTC (permalink / raw) To: Richard Henderson; +Cc: qemu-arm, QEMU Developers, Laurent Vivier On Fri, 18 Jun 2021 at 20:33, Richard Henderson <richard.henderson@linaro.org> wrote: > > ARM is more complicated than the others, in that we also > have trampolines for using SA_RESTORER with FDPIC, and > we need to create trampolines for both ARM and Thumb modes. > > Cc: qemu-arm@nongnu.org > Signed-off-by: Richard Henderson <richard.henderson@linaro.org> > --- > linux-user/arm/target_signal.h | 2 + > linux-user/arm/signal.c | 170 +++++++++++++++++++-------------- > 2 files changed, 100 insertions(+), 72 deletions(-) Reviewed-by: Peter Maydell <peter.maydell@linaro.org> though I'm not 100% confident I checked all the details. Do you have test programs that checked all the different trampolines ? thanks -- PMM ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH v2 05/23] linux-user/arm: Implement setup_sigtramp 2021-06-29 14:09 ` Peter Maydell @ 2021-06-29 18:32 ` Richard Henderson 0 siblings, 0 replies; 11+ messages in thread From: Richard Henderson @ 2021-06-29 18:32 UTC (permalink / raw) To: Peter Maydell; +Cc: qemu-arm, QEMU Developers, Laurent Vivier On 6/29/21 7:09 AM, Peter Maydell wrote: > On Fri, 18 Jun 2021 at 20:33, Richard Henderson > <richard.henderson@linaro.org> wrote: >> >> ARM is more complicated than the others, in that we also >> have trampolines for using SA_RESTORER with FDPIC, and >> we need to create trampolines for both ARM and Thumb modes. >> >> Cc: qemu-arm@nongnu.org >> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> >> --- >> linux-user/arm/target_signal.h | 2 + >> linux-user/arm/signal.c | 170 +++++++++++++++++++-------------- >> 2 files changed, 100 insertions(+), 72 deletions(-) > > Reviewed-by: Peter Maydell <peter.maydell@linaro.org> > > though I'm not 100% confident I checked all the details. Do you have > test programs that checked all the different trampolines ? No, I have no fdpic testcases, and wasn't keen to spend an afternoon building the toolchain. r~ ^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2021-07-01 19:27 UTC | newest] Thread overview: 11+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- [not found] <20210618192951.125651-1-richard.henderson@linaro.org> 2021-06-18 19:29 ` [PATCH v2 02/23] linux-user/aarch64: Implement setup_sigtramp Richard Henderson 2021-06-29 13:36 ` Peter Maydell 2021-07-01 19:27 ` Richard Henderson 2021-06-18 19:29 ` [PATCH v2 03/23] linux-user/arm: Split out v2_frame Richard Henderson 2021-06-29 13:53 ` Peter Maydell 2021-06-29 14:30 ` Richard Henderson 2021-06-18 19:29 ` [PATCH v2 04/23] linux-user/arm: Force v2 frames for fdpic Richard Henderson 2021-06-29 13:54 ` Peter Maydell 2021-06-18 19:29 ` [PATCH v2 05/23] linux-user/arm: Implement setup_sigtramp Richard Henderson 2021-06-29 14:09 ` Peter Maydell 2021-06-29 18:32 ` Richard Henderson
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).