bsd-user: Misc changes for 9.1 (I hope)
cover bsd-user: Misc changes for 9.1 (I hope)
This patch series includes two main sets of patches. To make it simple to review, I've included the changes from my student which the later changes depend on. I've included a change from Jessica and Doug as well. I've reviewed them, but more eyes never hurt.
I've also included a number of 'touch up' patches needed either to get the aarch64 building, or to implmement suggestions from prior review cycles. The main one is what's charitably described as a kludge: force aarch64 to use 4k pages. The qemu-project (and blitz branch) hasn't had the necessary changes to bsd-user needed to support variable page size.
Sorry this is so late... Live has conspired to delay me.
Doug Rabson (1):
bsd-user: Simplify the implementation of execve
Jessica Clarke (1):
bsd-user: Sync fork_start/fork_end with linux-user
Stacey Son (6):
bsd-user:Add CPU initialization and management functions bsd-user:Add AArch64 register handling and related functions bsd-user:Add ARM AArch64 signal handling support bsd-user:Add get_mcontext function for ARM AArch64 bsd-user:Add set_mcontext function for ARM AArch64 bsd-user:Add AArch64 improvements and signal handling functions
Warner Losh (6):
bsd-user:Add ARM AArch64 support and capabilities bsd-user:Add setup_sigframe_arch function for ARM AArch64
bsd-user: Hard wire aarch64 to be 4k pages only
bsd-user: Define TARGET_SIGSTACK_ALIGN and use it to round stack
bsd-user: Make compile for non-linux user-mode stuff
bsd-user: Add aarch64 build to tree
bsd-user/aarch64/signal.c | 137 +++++++++++++++++ bsd-user/aarch64/target_arch.h | 29 ++++ bsd-user/aarch64/target_arch_cpu.c | 31 ++++ bsd-user/aarch64/target_arch_cpu.h | 189 ++++++++++++++++++++++++ bsd-user/aarch64/target_arch_elf.h | 163 ++++++++++++++++++++ bsd-user/aarch64/target_arch_reg.h | 56 +++++++ bsd-user/aarch64/target_arch_signal.h | 82 ++++++++++ bsd-user/aarch64/target_arch_sigtramp.h | 48 ++++++ bsd-user/aarch64/target_arch_sysarch.h | 42 ++++++ bsd-user/aarch64/target_arch_thread.h | 61 ++++++++ bsd-user/aarch64/target_arch_vmparam.h | 74 ++++++++++ bsd-user/aarch64/target_syscall.h | 51 +++++++ bsd-user/arm/target_arch_signal.h | 2 + bsd-user/freebsd/os-proc.c | 118 +-------------- bsd-user/i386/target_arch_signal.h | 2 + bsd-user/main.c | 46 ++---- bsd-user/qemu.h | 3 + bsd-user/signal.c | 9 +- bsd-user/x86_64/target_arch_signal.h | 2 + configs/targets/aarch64-bsd-user.mak | 3 + target/arm/cpu-param.h | 4 + target/arm/gdbstub64.c | 4 + 22 files changed, 1001 insertions(+), 155 deletions(-) create mode 100644 bsd-user/aarch64/signal.c create mode 100644 bsd-user/aarch64/target_arch.h create mode 100644 bsd-user/aarch64/target_arch_cpu.c create mode 100644 bsd-user/aarch64/target_arch_cpu.h create mode 100644 bsd-user/aarch64/target_arch_elf.h create mode 100644 bsd-user/aarch64/target_arch_reg.h create mode 100644 bsd-user/aarch64/target_arch_signal.h create mode 100644 bsd-user/aarch64/target_arch_sigtramp.h create mode 100644 bsd-user/aarch64/target_arch_sysarch.h create mode 100644 bsd-user/aarch64/target_arch_thread.h create mode 100644 bsd-user/aarch64/target_arch_vmparam.h create mode 100644 bsd-user/aarch64/target_syscall.h create mode 100644 configs/targets/aarch64-bsd-user.mak
Signature · 1 line
2.45.1
PATCH 1/14 bsd-user:Add CPU initialization and management functions
From: Stacey Son <[email protected]>
Added function to initialize ARM CPU and check if it supports 64-bit mode. Implemented CPU loop function to handle exceptions and emulate execution of instructions. Added function to clone CPU state to create a new thread.
Included AArch64 specific CPU functions for bsd-user to set and receive thread-local-storage value from the tpidr_el0 register.
Introduced structure for storing CPU register states for BSD-USER.
Message-Id: <[email protected]>
--- bsd-user/aarch64/target_arch_cpu.c | 31 +++++ bsd-user/aarch64/target_arch_cpu.h | 192 +++++++++++++++++++++++++++++ bsd-user/aarch64/target_syscall.h | 51 ++++++++ 3 files changed, 274 insertions(+) create mode 100644 bsd-user/aarch64/target_arch_cpu.c create mode 100644 bsd-user/aarch64/target_arch_cpu.h create mode 100644 bsd-user/aarch64/target_syscall.h
bsd-user/aarch64/target_arch_cpu.c new file +31 -0
@@ -0,0 +1,31 @@ +/* +* ARM AArch64 specific CPU for bsd-user +* +* Copyright (c) 2015 Stacey Son +* +* This library is free software; you can redistribute it and/or +* modify it under the terms of the GNU Lesser General Public +* License as published by the Free Software Foundation; either +* version 2 of the License, or (at your option) any later version. +* +* This library is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +* Lesser General Public License for more details. +* +* You should have received a copy of the GNU Lesser General Public +* License along with this library; if not, see <http://www.gnu.org/licenses/>. +*/ +#include "qemu/osdep.h" +#include "target_arch.h" + +/* See cpu_set_user_tls() in arm64/arm64/vm_machdep.c */ +void target_cpu_set_tls (CPUARMState *env,target_ulong newtls) +{ + env->cp15.tpidr_el[0 ] = newtls; +} + +target_ulong target_cpu_get_tls (CPUARMState *env) +{ +return env->cp15.tpidr_el[0 ]; +}
bsd-user/aarch64/target_arch_cpu.h new file +192 -0
192 lines not on this page · raw message
bsd-user/aarch64/target_syscall.h new file +51 -0
51 lines not on this page · raw message
Signature · 1 line
2.45.1
PATCH 2/14 bsd-user:Add AArch64 register handling and related functions
In reply to Warner LoshFrom: Stacey Son <[email protected]>
Added header file for managing CPU register states in FreeBSD user mode.
Introduced prototypes for setting and getting thread-local storage (TLS).
Implemented AArch64 sysarch() system call emulation and a printing function.
Added function for setting up thread upcall to add thread support to BSD-USER. Initialized thread's register state during thread setup.
Updated ARM AArch64 VM parameter definitions for bsd-user, including address spaces for FreeBSD/arm64 and a function for getting the stack pointer from CPU and setting a return value.
Message-Id: <[email protected]>
--- bsd-user/aarch64/target_arch.h | 28 +++++++++++ bsd-user/aarch64/target_arch_reg.h | 56 +++++++++++++++++++++ bsd-user/aarch64/target_arch_sysarch.h | 42 ++++++++++++++++ bsd-user/aarch64/target_arch_thread.h | 61 +++++++++++++++++++++++ bsd-user/aarch64/target_arch_vmparam.h | 68 ++++++++++++++++++++++++++ 5 files changed, 255 insertions(+) create mode 100644 bsd-user/aarch64/target_arch.h create mode 100644 bsd-user/aarch64/target_arch_reg.h create mode 100644 bsd-user/aarch64/target_arch_sysarch.h create mode 100644 bsd-user/aarch64/target_arch_thread.h create mode 100644 bsd-user/aarch64/target_arch_vmparam.h
bsd-user/aarch64/target_arch.h new file +28 -0
@@ -0,0 +1,28 @@ +/* +* ARM AArch64 specific prototypes for bsd-user +* +* Copyright (c) 2015 Stacey D. Son <sson at FreeBSD> +* +* This library is free software; you can redistribute it and/or +* modify it under the terms of the GNU Lesser General Public +* License as published by the Free Software Foundation; either +* version 2 of the License, or (at your option) any later version. +* +* This library is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +* Lesser General Public License for more details. +* +* You should have received a copy of the GNU Lesser General Public +* License along with this library; if not, see <http://www.gnu.org/licenses/>. +*/ + +#ifndef TARGET_ARCH_H +#define TARGET_ARCH_H + +#include "qemu.h" + +void target_cpu_set_tls (CPUARMState *env,target_ulong newtls); +target_ulong target_cpu_get_tls (CPUARMState *env); + +#endif /* TARGET_ARCH_H */
bsd-user/aarch64/target_arch_reg.h new file +56 -0
56 lines not on this page · raw message
bsd-user/aarch64/target_arch_sysarch.h new file +42 -0
42 lines not on this page · raw message
bsd-user/aarch64/target_arch_thread.h new file +61 -0
61 lines not on this page · raw message
bsd-user/aarch64/target_arch_vmparam.h new file +68 -0
68 lines not on this page · raw message
Signature · 1 line
2.45.1
PATCH 3/14 bsd-user:Add ARM AArch64 support and capabilities
In reply to Warner LoshAdded function to access rval2 by accessing the x1 register.
Defined ARM AArch64 ELF parameters including mmap and dynamic load addresses.
Introduced extensive hardware capability definitions and macros for retrieving hardware capability (hwcap) flags. Implemented function to retrieve ARM AArch64 hardware capabilities using the GET_FEATURE_ID macro. Added function to retrieve extended ARM AArch64 hardware capability flags.
Message-Id: <[email protected]>
--- bsd-user/aarch64/target_arch.h | 1 + bsd-user/aarch64/target_arch_elf.h | 163 +++++++++++++++++++++++++ bsd-user/aarch64/target_arch_vmparam.h | 6 + 3 files changed, 170 insertions(+) create mode 100644 bsd-user/aarch64/target_arch_elf.h
bsd-user/aarch64/target_arch.h +1 -0
@@ -21,6 +21,7 @@#define TARGET_ARCH_H #include "qemu.h" +#include "target/arm/cpu-features.h" void target_cpu_set_tls (CPUARMState *env,target_ulong newtls);target_ulong target_cpu_get_tls (CPUARMState *env);
bsd-user/aarch64/target_arch_elf.h new file +163 -0
163 lines not on this page · raw message
bsd-user/aarch64/target_arch_vmparam.h +6 -0
@@ -65,4 +65,10 @@ static inline void set_second_rval(CPUARMState *state, abi_ulong retval2) { state->xregs[1 ] = retval2;/* XXX not really used on 64-bit arch */ } + +static inline abi_ulong get_second_rval (CPUARMState *state) +{ +return state->xregs[1 ]; +} +#endif /* TARGET_ARCH_VMPARAM_H */
Signature · 1 line
2.45.1
PATCH 4/14 bsd-user:Add ARM AArch64 signal handling support
In reply to Warner LoshFrom: Stacey Son <[email protected]>
Added sigcode setup function for signal trampoline which initializes a sequence of instructions to handle signal returns and exits, copying this code to the target offset.
Defined ARM AArch64 specific signal definitions including register indices and sizes,
and introduced structures to represent general purpose registers, floating point registers, and machine context. Added function to set up signal handler arguments, populating register values in CPUARMState based on the provided signal, signal frame, signal action, and frame address.
Message-Id: <[email protected]>
--- bsd-user/aarch64/signal.c | 53 ++++++++++++++++ bsd-user/aarch64/target_arch_signal.h | 80 +++++++++++++++++++++++++ bsd-user/aarch64/target_arch_sigtramp.h | 48 +++++++++++++++ 3 files changed, 181 insertions(+) create mode 100644 bsd-user/aarch64/signal.c create mode 100644 bsd-user/aarch64/target_arch_signal.h create mode 100644 bsd-user/aarch64/target_arch_sigtramp.h
bsd-user/aarch64/signal.c new file +53 -0
@@ -0,0 +1,53 @@ +/* +* ARM AArch64 specific signal definitions for bsd-user +* +* Copyright (c) 2015 Stacey D. Son <sson at FreeBSD> +* +* This library is free software; you can redistribute it and/or +* modify it under the terms of the GNU Lesser General Public +* License as published by the Free Software Foundation; either +* version 2 of the License, or (at your option) any later version. +* +* This library is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +* Lesser General Public License for more details. +* +* You should have received a copy of the GNU Lesser General Public +* License along with this library; if not, see <http://www.gnu.org/licenses/>. +*/ +#include "qemu/osdep.h" + +#include "qemu.h" + +/* +* Compare to sendsig() in sys/arm64/arm64/machdep.c +* Assumes that target stack frame memory is locked. +*/ +abi_long set_sigtramp_args (CPUARMState *regs,int sig, +struct target_sigframe *frame, +abi_ulong frame_addr, +struct target_sigaction *ka) +{ +/* +* Arguments to signal handler: +* x0 = signal number +* x1 = siginfo pointer +* x2 = ucontext pointer +* pc/elr = signal handler pointer +* sp = sigframe struct pointer +* lr = sigtramp at base of user stack +*/ + + regs->xregs[0 ] = sig; + regs->xregs[1 ] = frame_addr + + offsetof(struct target_sigframe , sf_si); + regs->xregs[2 ] = frame_addr + + offsetof(struct target_sigframe , sf_uc); + + regs->pc = ka->_sa_handler; + regs->xregs[TARGET_REG_SP ] = frame_addr; + regs->xregs[TARGET_REG_LR ] =TARGET_PS_STRINGS -TARGET_SZSIGCODE ; + +return 0 ; +}
bsd-user/aarch64/target_arch_signal.h new file +80 -0
80 lines not on this page · raw message
bsd-user/aarch64/target_arch_sigtramp.h new file +48 -0
48 lines not on this page · raw message
Signature · 1 line
2.45.1
PATCH 5/14 bsd-user:Add get_mcontext function for ARM AArch64
In reply to Warner LoshFrom: Stacey Son <[email protected]>
function to retrieve machine context,it populates the provided target_mcontext_t structure with information from the CPUARMState registers.
Message-Id: <[email protected]>
--- bsd-user/aarch64/signal.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+)
bsd-user/aarch64/signal.c +30 -0
@@ -51,3 +51,33 @@ abi_long set_sigtramp_args(CPUARMState *regs, int sig,return 0 ; } + +/* +* Compare to get_mcontext() in arm64/arm64/machdep.c +* Assumes that the memory is locked if mcp points to user memory. +*/ +abi_long get_mcontext (CPUARMState *regs,target_mcontext_t *mcp,int flags) +{ +int err =0 , i; +uint64_t *gr = mcp->mc_gpregs.gp_x; + + mcp->mc_gpregs.gp_spsr =pstate_read (regs); +if (flags &TARGET_MC_GET_CLEAR_RET ) { + gr[0 ] =0UL ; + mcp->mc_gpregs.gp_spsr &= ~CPSR_C ; + }else { + gr[0 ] =tswap64 (regs->xregs[0 ]); + } + +for (i =1 ; i <30 ; i++) { + gr[i] =tswap64 (regs->xregs[i]); + } + + mcp->mc_gpregs.gp_sp =tswap64 (regs->xregs[TARGET_REG_SP ]); + mcp->mc_gpregs.gp_lr =tswap64 (regs->xregs[TARGET_REG_LR ]); + mcp->mc_gpregs.gp_elr =tswap64 (regs->pc); + +/* XXX FP? */ + +return err; +}
Signature · 1 line
2.45.1
PATCH 6/14 bsd-user:Add setup_sigframe_arch function for ARM AArch64
In reply to Warner LoshThe function utilizes the get_mcontext function to retrieve the machine context for the current CPUARMState
Message-Id: <[email protected]>
--- bsd-user/aarch64/signal.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)
bsd-user/aarch64/signal.c +14 -0
@@ -81,3 +81,17 @@ abi_long get_mcontext(CPUARMState *regs, target_mcontext_t *mcp, int flags)return err; } + +/* +* Compare to arm64/arm64/exec_machdep.c sendsig() +* Assumes that the memory is locked if frame points to user memory. +*/ +abi_long setup_sigframe_arch (CPUARMState *env,abi_ulong frame_addr, +struct target_sigframe *frame,int flags) +{ +target_mcontext_t *mcp = &frame->sf_uc.uc_mcontext; + +get_mcontext (env, mcp, flags); +return 0 ; +} +
Signature · 1 line
2.45.1
PATCH 7/14 bsd-user:Add set_mcontext function for ARM AArch64
In reply to Warner LoshFrom: Stacey Son <[email protected]>
The function copies register values from the provided target_mcontext_t structure to the CPUARMState registers.
Note:FP is unfinished upstream but will be a separate commit coming soon.
Message-Id: <[email protected]>
--- bsd-user/aarch64/signal.c | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+)
bsd-user/aarch64/signal.c +22 -0
@@ -95,3 +95,25 @@ abi_long setup_sigframe_arch(CPUARMState *env, abi_ulong frame_addr,return 0 ; } +/* +* Compare to set_mcontext() in arm64/arm64/machdep.c +* Assumes that the memory is locked if frame points to user memory. +*/ +abi_long set_mcontext (CPUARMState *regs,target_mcontext_t *mcp,int srflag) +{ +int err =0 , i; +const uint64_t *gr = mcp->mc_gpregs.gp_x; + +for (i =0 ; i <30 ; i++) { + regs->xregs[i] =tswap64 (gr[i]); + } + + regs->xregs[TARGET_REG_SP ] =tswap64 (mcp->mc_gpregs.gp_sp); + regs->xregs[TARGET_REG_LR ] =tswap64 (mcp->mc_gpregs.gp_lr); + regs->pc = mcp->mc_gpregs.gp_elr; +pstate_write (regs, mcp->mc_gpregs.gp_spsr); + +/* XXX FP? */ + +return err; +}
Signature · 1 line
2.45.1
PATCH 8/14 bsd-user:Add AArch64 improvements and signal handling functions
In reply to Warner LoshFrom: Stacey Son <[email protected]>
Added get_ucontext_sigreturn function to check processor state ensuring current execution mode is EL0 and no flags indicating interrupts or exceptions are set.
Updated AArch64 code to use CF directly without reading/writing the entire processor state, improving efficiency. Changed FP data structures to use Int128 instead of __uint128_t, leveraging QEMU's generic mechanism for referencing this type.
Message-Id: <[email protected]>
--- bsd-user/aarch64/signal.c | 20 +++++++++++++++++++- bsd-user/aarch64/target_arch_cpu.h | 7 ++----- bsd-user/aarch64/target_arch_reg.h | 2 +- bsd-user/aarch64/target_arch_signal.h | 2 +- bsd-user/qemu.h | 3 +++ 5 files changed, 26 insertions(+), 8 deletions(-)
bsd-user/aarch64/signal.c +19 -1
@@ -21,7 +21,7 @@#include "qemu.h" /* -* Compare to sendsig() in sys/arm64/arm64/machdep.c +* Compare to sendsig() in sys/arm64/arm64/exec_machdep.c * Assumes that target stack frame memory is locked. */ abi_long set_sigtramp_args(CPUARMState *regs, int sig,
@@ -117,3 +117,21 @@ abi_long set_mcontext(CPUARMState *regs, target_mcontext_t *mcp, int srflag)return err; } + +/* Compare to sys_sigreturn() in arm64/arm64/machdep.c */ +abi_long get_ucontext_sigreturn (CPUARMState *regs,abi_ulong target_sf, +abi_ulong *target_uc) +{ +uint32_t pstate =pstate_read (regs); + + *target_uc =0 ; + +if ((pstate &PSTATE_M ) != PSTATE_MODE_EL0t || + (pstate & (PSTATE_F |PSTATE_I |PSTATE_A |PSTATE_D )) !=0 ) { +return -TARGET_EINVAL ; + } + + *target_uc = target_sf; + +return 0 ; +}
bsd-user/aarch64/target_arch_cpu.h +2 -5
@@ -48,7 +48,6 @@ static inline void target_cpu_loop(CPUARMState *env)CPUState *cs =env_cpu (env);int trapnr, ec, fsc, si_code, si_signo;uint64_t code, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8; -uint32_t pstate;abi_long ret;for (;;) {
@@ -88,18 +87,16 @@ static inline void target_cpu_loop(CPUARMState *env)* The carry bit is cleared for no error; set for error. * See arm64/arm64/vm_machdep.c cpu_set_syscall_retval() */ - pstate =pstate_read (env);if (ret >=0 ) { - pstate &= ~PSTATE_C ; + env->CF =0 ; env->xregs[0 ] = ret; }else if (ret == -TARGET_ERESTART ) { env->pc -=4 ;break ; }else if (ret != -TARGET_EJUSTRETURN ) { - pstate |=PSTATE_C ; + env->CF =1 ; env->xregs[0 ] = -ret; } -pstate_write (env, pstate);break ;case EXCP_INTERRUPT :
bsd-user/aarch64/target_arch_reg.h +1 -1
@@ -31,7 +31,7 @@ typedef struct target_reg { } target_reg_t;typedef struct target_fpreg { -__uint128_t fp_q[32 ]; +Int128 fp_q[32 ];uint32_t fp_sr;uint32_t fp_cr; }target_fpreg_t ;
bsd-user/aarch64/target_arch_signal.h +1 -1
@@ -49,7 +49,7 @@ struct target_gpregs { };struct target_fpregs { -__uint128_t fp_q[32 ]; +Int128 fp_q[32 ];uint32_t fp_sr;uint32_t fp_cr;uint32_t fp_flags;
bsd-user/qemu.h +3 -0
@@ -17,6 +17,9 @@#ifndef QEMU_H #define QEMU_H +#include <sys/param.h> + +#include "qemu/int128.h" #include "cpu.h" #include "qemu/units.h" #include "exec/cpu_ldst.h"
Signature · 1 line
2.45.1
PATCH 9/14 bsd-user: Simplify the implementation of execve
In reply to Warner LoshFrom: Doug Rabson <[email protected]>
This removes the logic which prepends the emulator to each call to execve and fexecve. This is not necessary with the existing imgact_binmisc support and it avoids the need to install the emulator binary into jail environments when using 'binmiscctl --pre-open'.
--- bsd-user/freebsd/os-proc.c | 118 +------------------------------------ bsd-user/main.c | 18 ------ 2 files changed, 3 insertions(+), 133 deletions(-)
bsd-user/freebsd/os-proc.c +3 -115
@@ -26,65 +26,13 @@ struct kinfo_proc;#include "qemu.h" -/* -* Get the filename for the given file descriptor. -* Note that this may return NULL (fail) if no longer cached in the kernel. -*/ -static char * -get_filename_from_fd (pid_t pid,int fd,char *filename,size_t len) -{ -char *ret =NULL ; -unsigned int cnt; -struct procstat *procstat =NULL ; -struct kinfo_proc *kp =NULL ; -struct filestat_list *head =NULL ; -struct filestat *fst; - - procstat =procstat_open_sysctl (); -if (procstat ==NULL ) { - goto out; - } - - kp =procstat_getprocs (procstat,KERN_PROC_PID , pid, &cnt); -if (kp ==NULL ) { - goto out; - } - - head =procstat_getfiles (procstat, kp,0 ); -if (head ==NULL ) { - goto out; - } - - STAILQ_FOREACH(fst, head, next) { -if (fd == fst->fs_fd) { -if (fst->fs_path !=NULL ) { - (void )strlcpy (filename, fst->fs_path, len); - ret = filename; - } -break ; - } - } - -out: -if (head !=NULL ) { -procstat_freefiles (procstat, head); - } -if (kp !=NULL ) { -procstat_freeprocs (procstat, kp); - } -if (procstat !=NULL ) { -procstat_close (procstat); - } -return ret; -} -/* * execve/fexecve */ abi_long freebsd_exec_common (abi_ulong path_or_fd,abi_ulong guest_argp,abi_ulong guest_envp,int do_fexec) { -char **argp, **envp, **qargp, **qarg1, **qarg0, **qargend; +char **argp, **envp, **qarg0;int argc, envc;abi_ulong gp;abi_ulong addr;
@@ -117,9 +65,7 @@ abi_long freebsd_exec_common(abi_ulong path_or_fd, abi_ulong guest_argp, qarg0 = argp = g_new0(char *, argc + 9);/* save the first argument for the emulator */ *argp++ = (char *)getprogname (); - qargp = argp; *argp++ = (char *)getprogname (); - qarg1 = argp; envp = g_new0(char *, envc + 1);for (gp = guest_argp, q = argp; gp; gp +=sizeof (abi_ulong), q++) {if (get_user_ual (addr, gp)) {
@@ -137,7 +83,6 @@ abi_long freebsd_exec_common(abi_ulong path_or_fd, abi_ulong guest_argp, total_size +=strlen (*q) +1 ; } *q++ =NULL ; - qargend = q;for (gp = guest_envp, q = envp; gp; gp +=sizeof (abi_ulong), q++) {if (get_user_ual (addr, gp)) {
@@ -166,71 +111,14 @@ abi_long freebsd_exec_common(abi_ulong path_or_fd, abi_ulong guest_argp, }if (do_fexec) { -if (((int )path_or_fd >0 && -is_target_elf_binary ((int )path_or_fd)) ==1 ) { -char execpath[PATH_MAX ]; - -/* -* The executable is an elf binary for the target -* arch. execve() it using the emulator if we can -* determine the filename path from the fd. -*/ -if (get_filename_from_fd (getpid (), (int )path_or_fd, execpath, -sizeof (execpath)) !=NULL ) { -memmove (qarg1 +2 , qarg1, (qargend - qarg1) *sizeof (*qarg1)); - qarg1[1 ] = qarg1[0 ]; - qarg1[0 ] = (char *)"-0" ; - qarg1 +=2 ; - qargend +=2 ; - *qarg1 = execpath; -#ifndef DONT_INHERIT_INTERP_PREFIX -memmove (qarg1 +2 , qarg1, (qargend - qarg1) *sizeof (*qarg1)); - *qarg1++ = (char *)"-L" ; - *qarg1++ = (char *)interp_prefix; -#endif - ret =get_errno (execve (qemu_proc_pathname, qargp, envp)); - }else { -/* Getting the filename path failed. */ - ret = -TARGET_EBADF ; - goto execve_end; - } - }else { - ret =get_errno (fexecve ((int )path_or_fd, argp, envp)); - } + ret =get_errno (fexecve ((int )path_or_fd, argp, envp)); }else { -int fd; - p =lock_user_string (path_or_fd);if (p ==NULL ) { ret = -TARGET_EFAULT ; goto execve_end; } - -/* -* Check the header and see if it a target elf binary. If so -* then execute using qemu user mode emulator. -*/ - fd =open (p,O_RDONLY |O_CLOEXEC ); -if (fd >0 &&is_target_elf_binary (fd) ==1 ) { -close (fd); -/* execve() as a target binary using emulator. */ -memmove (qarg1 +2 , qarg1, (qargend - qarg1) *sizeof (*qarg1)); - qarg1[1 ] = qarg1[0 ]; - qarg1[0 ] = (char *)"-0" ; - qarg1 +=2 ; - qargend +=2 ; - *qarg1 = (char *)p; -#ifndef DONT_INHERIT_INTERP_PREFIX -memmove (qarg1 +2 , qarg1, (qargend - qarg1) *sizeof (*qarg1)); - *qarg1++ = (char *)"-L" ; - *qarg1++ = (char *)interp_prefix; -#endif - ret =get_errno (execve (qemu_proc_pathname, qargp, envp)); - }else { -close (fd); -/* Execve() as a host native binary. */ - ret =get_errno (execve (p, argp, envp)); - } + ret =get_errno (execve (p, argp, envp));unlock_user (p, path_or_fd,0 ); }
bsd-user/main.c +0 -18
@@ -90,7 +90,6 @@ unsigned long reserved_va;const char *interp_prefix =CONFIG_QEMU_INTERP_PREFIX ;const char *qemu_uname_release; -char qemu_proc_pathname[PATH_MAX ];/* full path to exeutable */ unsigned long target_maxtsiz =TARGET_MAXTSIZ ;/* max text size */ unsigned long target_dfldsiz =TARGET_DFLDSIZ ;/* initial data size limit */
@@ -247,22 +246,6 @@ adjust_ssize(void)setrlimit (RLIMIT_STACK , &rl); } -static void save_proc_pathname (char *argv0) -{ -int mib[4 ]; -size_t len; - - mib[0 ] =CTL_KERN ; - mib[1 ] =KERN_PROC ; - mib[2 ] =KERN_PROC_PATHNAME ; - mib[3 ] =-1 ; - - len =sizeof (qemu_proc_pathname); -if (sysctl (mib,4 , qemu_proc_pathname, &len,NULL ,0 )) { -perror ("sysctl" ); - } -} -int main (int argc,char **argv) {const char *filename;
@@ -292,7 +275,6 @@ int main(int argc, char **argv)usage (); } -save_proc_pathname (argv[0 ]);error_init (argv[0 ]);module_call_init (MODULE_INIT_TRACE );
Signature · 1 line
2.45.1
Re: [PATCH 09/14] bsd-user: Simplify the implementation of execve
Quoted 7 lines · Warner Losh
From: Doug Rabson<[email protected]>
This removes the logic which prepends the emulator to each call to execve and fexecve. This is not necessary with the existing
imgact_binmiscsupport and it avoids the need to install the emulator binary into jail environments when using 'binmiscctl --pre-open'.
Quoted 7 lines
Signed-off-by Doug Rabson<[email protected]>Reviewed-by Warner Losh<[email protected]>Signed-off-by Warner Losh<[email protected]>--- bsd-user/freebsd/os-proc.c | 118 +------------------------------------ bsd-user/main.c | 18 ------ 2 files changed, 3 insertions(+), 133 deletions(-)
r~
PATCH 10/14 bsd-user: Hard wire aarch64 to be 4k pages only
In reply to Warner LoshOnly support 4k pages for aarch64 binaries. The variable page size stuff isn't working just yet, so put in this lessor-of-evils kludge until that is complete.
--- target/arm/cpu-param.h | 4 ++++ 1 file changed, 4 insertions(+)
target/arm/cpu-param.h +4 -0
@@ -21,9 +21,13 @@#ifdef CONFIG_USER_ONLY # ifdef TARGET_AARCH64 # define TARGET_TAGGED_ADDRESSES +# ifdef __FreeBSD__ +# define TARGET_PAGE_BITS 12 +# else /* Allow user-only to vary page size from 4k */ # define TARGET_PAGE_BITS_VARY # define TARGET_PAGE_BITS_MIN 12 +# endif # else # define TARGET_PAGE_BITS 12# endif
Signature · 1 line
2.45.1
Re: [PATCH 10/14] bsd-user: Hard wire aarch64 to be 4k pages only
Quoted 3 lines · Warner Losh
Only support 4k pages for aarch64 binaries. The variable page size stuff isn't working just yet, so put in this lessor-of-evils kludge until that is complete.
Hmm. In what way is it not working?
What if we limit this to bsd-user/main.c instead?
set_preferred_target_page_bits(TARGET_PAGE_BITS_MIN);
r~
Re: [PATCH 10/14] bsd-user: Hard wire aarch64 to be 4k pages only
Quoted 11 lines · Richard Henderson
Quoted 3 lines · Warner Losh
Only support 4k pages for aarch64 binaries. The variable page size stuff isn't working just yet, so put in this lessor-of-evils kludge until that is complete.
Hmm. In what way is it not working?
What if we limit this to
bsd-user/main.cinstead?set_preferred_target_page_bits(TARGET_PAGE_BITS_MIN);
It's a compile issue that this won't solve:
In file included from ../bsd-user/aarch64/signal.c:21:../bsd-user/qemu.h:146:20: warning: variable length array used [-Wvla]
146 | void *page[MAX_ARG_PAGES];
| ^~~~~~~~~~~~~
../bsd-user/qemu.h:138:23: note: expanded from macro 'MAX_ARG_PAGES'
138 | #define MAX_ARG_PAGES (TARGET_ARG_MAX / TARGET_PAGE_SIZE)
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~../bsd-user/qemu.h:146:20: note: initializer of 'target_page' is unknown ../bsd-user/qemu.h:138:41: note: expanded from macro 'MAX_ARG_PAGES'
138 | #define MAX_ARG_PAGES (TARGET_ARG_MAX / TARGET_PAGE_SIZE)
| ^
/dune/imp/git/qemu/include/exec/cpu-all.h:152:37: note: expanded from macro 'TARGET_PAGE_SIZE'
152 | # define TARGET_PAGE_SIZE (-(int)TARGET_PAGE_MASK)
| ^
/dune/imp/git/qemu/include/exec/cpu-all.h:150:44: note: expanded from macro 'TARGET_PAGE_MASK'
150 | # define TARGET_PAGE_MASK ((target_long)target_page.mask)
| ^
/dune/imp/git/qemu/include/exec/cpu-all.h:142:29: note: declared here
142 | extern const TargetPageBits target_page;
| ^
In file included from ../bsd-user/aarch64/signal.c:21:../bsd-user/qemu.h:146:15: error: fields must have a constant size:
'variable length array in structure' extension will never be supported
146 | void *page[MAX_ARG_PAGES];
| ^
1 warning and 1 error generated.
ninja: build stopped: subcommand failed.
gmake: *** [Makefile:167: run-ninja] Error 1
that have a couple of additional levels after those issues are fixed... It's fixable, but will take a fair amount more time that I have before the deadline...
Warner
Richard Henderson
Reviewed-byRe: [PATCH 10/14] bsd-user: Hard wire aarch64 to be 4k pages only
Quoted 1 line · Warner Losh
It's a compile issue that this won't solve:
...
Quoted 1 line
It's fixable, but will take a fair amount more time that I have before the deadline...
Ok, fair.
r~
PATCH 11/14 bsd-user: Sync fork_start/fork_end with linux-user
In reply to Warner LoshFrom: Jessica Clarke <[email protected]>
This reorders some of the calls, deduplicates code between branches and, most importantly, fixes a double end_exclusive call in the parent that will cause exclusive_context_count to go negative.
Pull-Request: https://github.com/qemu-bsd-user/qemu-bsd-user/pull/52
--- bsd-user/main.c | 28 ++++++++++++++-------------- 1 file changed, 14 insertions(+), 14 deletions(-)
bsd-user/main.c +14 -14
@@ -35,6 +35,7 @@#include "qemu/path.h" #include "qemu/help_option.h" #include "qemu/module.h" +#include "qemu/plugin.h" #include "exec/exec-all.h" #include "user/guest-base.h" #include "tcg/startup.h"
@@ -103,8 +104,9 @@ unsigned long target_sgrowsiz = TARGET_SGROWSIZ; /* amount to grow stack */void fork_start (void ) {start_exclusive (); -cpu_list_lock ();mmap_fork_start (); +cpu_list_lock (); +qemu_plugin_user_prefork_lock ();gdbserver_fork_start (); }
@@ -112,31 +114,29 @@ void fork_end(pid_t pid) {bool child = pid ==0 ; +qemu_plugin_user_postfork (child); +mmap_fork_end (child);if (child) {CPUState *cpu, *next_cpu; -/* -* Child processes created by fork() only have a single thread. Discard -* information about the parent threads. -*/ +/* Child processes created by fork() only have a single thread. +Discard information about the parent threads. */ CPU_FOREACH_SAFE(cpu, next_cpu) {if (cpu != thread_cpu) {QTAILQ_REMOVE_RCU (&cpus_queue, cpu, node); } } -mmap_fork_end (child); -/* -* qemu_init_cpu_list() takes care of reinitializing the exclusive -* state, so we don't need to end_exclusive() here. -*/ qemu_init_cpu_list ();get_task_state (thread_cpu)->ts_tid =qemu_get_thread_id (); -gdbserver_fork_end (thread_cpu, pid); }else { -mmap_fork_end (child);cpu_list_unlock (); -gdbserver_fork_end (thread_cpu, pid); -end_exclusive (); } +gdbserver_fork_end (thread_cpu, pid); +/* +* qemu_init_cpu_list() reinitialized the child exclusive state, but we +* also need to keep current_cpu consistent, so call end_exclusive() for +* both child and parent. +*/ +end_exclusive (); }void cpu_loop (CPUArchState *env)
Signature · 1 line
2.45.1
Re: [PATCH 11/14] bsd-user: Sync fork_start/fork_end with linux-user
bsd-user/main.c void fork_end(pid_t pid) · line 122
Quoted 6 lines · Warner Losh
bsd-user/main.c +2
-4-/* -* Child processes created by fork() only have a single thread. Discard -* information about the parent threads. -*/ +/* Child processes created by fork() only have a single thread. +Discard information about the parent threads. */
This comment should stay formatted the way it is, per coding style.
Otherwise,
r~
PATCH 12/14 bsd-user: Define TARGET_SIGSTACK_ALIGN and use it to round stack
In reply to Warner LoshMost (all?) targets require stacks to be properly aligned. Rather than a series of ifdefs in bsd-user/signal.h, instead use a manditory #define for all architectures.
--- bsd-user/aarch64/target_arch_signal.h | 2 ++ bsd-user/arm/target_arch_signal.h | 2 ++ bsd-user/i386/target_arch_signal.h | 2 ++ bsd-user/signal.c | 9 +-------- bsd-user/x86_64/target_arch_signal.h | 2 ++ 5 files changed, 9 insertions(+), 8 deletions(-)
bsd-user/aarch64/target_arch_signal.h +2 -0
@@ -77,4 +77,6 @@ struct target_sigframe { target_ucontext_t sf_uc; /* saved ucontext */ }; +#define TARGET_SIGSTACK_ALIGN 16 +#endif /* TARGET_ARCH_SIGNAL_H */
bsd-user/arm/target_arch_signal.h +2 -0
@@ -86,4 +86,6 @@ struct target_sigframe { target_mcontext_vfp_t sf_vfp; /* actual saved VFP context */ }; +#define TARGET_SIGSTACK_ALIGN 8 +#endif /* TARGET_ARCH_SIGNAL_H */
bsd-user/i386/target_arch_signal.h +2 -0
@@ -88,4 +88,6 @@ struct target_sigframe { uint32_t __spare__[2]; }; +#define TARGET_SIGSTACK_ALIGN 8 +#endif /* TARGET_ARCH_SIGNAL_H */
bsd-user/signal.c +1 -8
@@ -728,14 +728,7 @@ static inline abi_ulong get_sigframe(struct target_sigaction *ka, sp = ts->sigaltstack_used.ss_sp + ts->sigaltstack_used.ss_size; } -/* TODO: make this a target_arch function / define */ -#if defined(TARGET_ARM ) -return (sp - frame_size) & ~7 ; -#elif defined(TARGET_AARCH64 ) -return (sp - frame_size) & ~15 ; -#else -return sp - frame_size; -#endif +return ROUND_DOWN (sp - frame_size,TARGET_SIGSTACK_ALIGN ); }/* compare to $M/$M/exec_machdep.c sendsig and sys/kern/kern_sig.c sigexit */
bsd-user/x86_64/target_arch_signal.h +2 -0
@@ -97,4 +97,6 @@ struct target_sigframe { uint32_t __spare__[2]; }; +#define TARGET_SIGSTACK_ALIGN 16 +#endif /* TARGET_ARCH_SIGNAL_H */
Signature · 1 line
2.45.1
Re: [PATCH 12/14] bsd-user: Define TARGET_SIGSTACK_ALIGN and use it to round stack
Quoted 4 lines · Warner Losh
Most (all?) targets require stacks to be properly aligned. Rather than a series of ifdefs in
bsd-user/signal.h, instead use a manditory #define for all architectures.
Quoted 8 lines
Signed-off-by Warner Losh<[email protected]>--- bsd-user/aarch64/target_arch_signal.h | 2 ++ bsd-user/arm/target_arch_signal.h | 2 ++ bsd-user/i386/target_arch_signal.h | 2 ++ bsd-user/signal.c | 9 +-------- bsd-user/x86_64/target_arch_signal.h | 2 ++ 5 files changed, 9 insertions(+), 8 deletions(-)
r~
PATCH 13/14 bsd-user: Make compile for non-linux user-mode stuff
In reply to Warner LoshWe include the files that define PR_MTE_TCF_SHIFT only on Linux, but use them unconditionally. Restrict its use to Linux-only.
--- target/arm/gdbstub64.c | 4 ++++ 1 file changed, 4 insertions(+)
target/arm/gdbstub64.c +4 -0
@@ -404,6 +404,7 @@ int aarch64_gdb_get_tag_ctl_reg(CPUState *cs, GByteArray *buf, int reg)int aarch64_gdb_set_tag_ctl_reg (CPUState *cs,uint8_t *buf,int reg) { +#if defined(CONFIG_LINUX )ARMCPU *cpu =ARM_CPU (cs);CPUARMState *env = &cpu->env;
@@ -425,6 +426,9 @@ int aarch64_gdb_set_tag_ctl_reg(CPUState *cs, uint8_t *buf, int reg)arm_set_mte_tcf0 (env, tcf);return 1 ; +#else +return 0 ; +#endif }static void handle_q_memtag (GArray *params,void *user_ctx)
Signature · 1 line
2.45.1
Re: [PATCH 13/14] bsd-user: Make compile for non-linux user-mode stuff
Hi Warner,
Quoted 2 lines · Warner Losh
We include the files that define
PR_MTE_TCF_SHIFTonly on Linux, but use them unconditionally. Restrict its use to Linux-only.
We should check that in meson, i.e.:
config_host_data.set('CONFIG_PRCTL_PR_MTE_TCF_SHIFT',
cc.has_header_symbol('sys/prctl.h', 'PR_MTE_TCF_SHIFT'))
(like we do for CONFIG_PRCTL_PR_SET_TIMERSLACK), then rework linux-user/aarch64/mte_user_helper.h (and possibly tests/tcg/aarch64/mte.h), moving in a common directory.
That said, your patch matches this file header:
#if defined(CONFIG_USER_ONLY) && defined(CONFIG_LINUX) #include <sys/prctl.h> #include "mte_user_helper.h" #endif
Even if this #ifdef'ry is ugly, it isn't wrong, so:
¯\_(ツ)_/¯
Quoted 27 lines
Signed-off-by Warner Losh <[email protected]>--- target/arm/gdbstub64.c | 4 ++++ 1 file changed, 4 insertions(+)target/arm/gdbstub64.c +4
-0@@ -404,6 +404,7 @@ int aarch64_gdb_get_tag_ctl_reg(CPUState *cs, GByteArray *buf, int reg)int aarch64_gdb_set_tag_ctl_reg (CPUState *cs,uint8_t *buf,int reg) { +#if defined(CONFIG_LINUX )ARMCPU *cpu =ARM_CPU (cs);CPUARMState *env = &cpu->env;@@ -425,6 +426,9 @@ int aarch64_gdb_set_tag_ctl_reg(CPUState *cs, uint8_t *buf, int reg)arm_set_mte_tcf0 (env, tcf);return 1 ; +#else +return 0 ; +#endif }static void handle_q_memtag (GArray *params,void *user_ctx)
Re: [PATCH 13/14] bsd-user: Make compile for non-linux user-mode stuff
Quoted 11 lines · Philippe Mathieu-Daudé
Hi Warner,
Quoted 2 lines · Warner Losh
We include the files that define
PR_MTE_TCF_SHIFTonly on Linux, but use them unconditionally. Restrict its use to Linux-only.We should check that in meson, i.e.:
config_host_data.set('CONFIG_PRCTL_PR_MTE_TCF_SHIFT',
cc.has_header_symbol('sys/prctl.h', 'PR_MTE_TCF_SHIFT'))
Surely this fails for non-aarch64 hosts?
Quoted 40 lines
(like we do for
CONFIG_PRCTL_PR_SET_TIMERSLACK), then reworklinux-user/aarch64/mte_user_helper.h(and possiblytests/tcg/aarch64/mte.h), moving in a common directory.That said, your patch matches this file header:
#if defined(CONFIG_USER_ONLY) && defined(CONFIG_LINUX) #include <sys/prctl.h> #include "mte_user_helper.h" #endifEven if this #ifdef'ry is ugly, it isn't wrong, so:
Reviewed-by Philippe Mathieu-Daudé <[email protected]>¯\_(ツ)_/¯
Quoted 24 lines
Signed-off-by Warner Losh <[email protected]>--- target/arm/gdbstub64.c | 4 ++++ 1 file changed, 4 insertions(+)target/arm/gdbstub64.c +1
-0@@ -404,6 +404,7 @@ int aarch64_gdb_get_tag_ctl_reg(CPUState *cs, GByteArray *buf, int reg)int aarch64_gdb_set_tag_ctl_reg (CPUState *cs,uint8_t *buf,int reg) { +#if defined(CONFIG_LINUX )ARMCPU *cpu =ARM_CPU (cs);CPUARMState *env = &cpu->env; @@ -425,6 +426,9 @@ int aarch64_gdb_set_tag_ctl_reg(CPUState *cs,
uint8_t*buf, int reg)arm_set_mte_tcf0(env, tcf); return 1;target/arm/gdbstub64.c +3
-0+#else +return 0 ; +#endif }static void handle_q_memtag (GArray *params,void *user_ctx)
Signature · 2 lines
Alex Bennée Virtualisation Tech Lead @ Linaro
Re: [PATCH 13/14] bsd-user: Make compile for non-linux user-mode stuff
In reply to Philippe Mathieu-DaudéQuoted 8 lines · Philippe Mathieu-Daudé
Hi Warner,
Quoted 2 lines · Warner Losh
We include the files that define
PR_MTE_TCF_SHIFTonly on Linux, but use them unconditionally. Restrict its use to Linux-only.We should check that in meson, i.e.:
Quoted 8 lines
config_host_data.set('CONFIG_PRCTL_PR_MTE_TCF_SHIFT',
cc.has_header_symbol('sys/prctl.h','
PR_MTE_TCF_SHIFT'))(like we do for
CONFIG_PRCTL_PR_SET_TIMERSLACK), then reworklinux-user/aarch64/mte_user_helper.h(and possiblytests/tcg/aarch64/mte.h), moving in a common directory.
Hmmm... I'll have to check, but since I didn't add the incompatibility, it will likely be on my backlog for a while...
Quoted 7 lines
That said, your patch matches this file header:
#if defined(CONFIG_USER_ONLY) && defined(CONFIG_LINUX) #include <sys/prctl.h> #include "mte_user_helper.h" #endif
Quoted 5 lines
Even if this #ifdef'ry is ugly, it isn't wrong, so:
Reviewed-by Philippe Mathieu-Daudé <[email protected]>¯\_(ツ)_/¯
Thanks!
Warner
Quoted 31 lines
target/arm/gdbstub64.c int aarch64_gdb_get_tag_ctl_reg(CPUState *cs,
Quoted 10 lines
Signed-off-by Warner Losh <[email protected]>--- target/arm/gdbstub64.c | 4 ++++ 1 file changed, 4 insertions(+)target/arm/gdbstub64.c +0
-0@@ -404,6 +404,7 @@ int aarch64_gdb_get_tag_ctl_reg(CPUState *cs,GByteArray *buf, int reg)
target/arm/gdbstub64.c int aarch64_gdb_set_tag_ctl_reg(CPUState *cs,
Quoted 8 lines
target/arm/gdbstub64.c +1
-0int aarch64_gdb_set_tag_ctl_reg (CPUState *cs,uint8_t *buf,int reg) { +#if defined(CONFIG_LINUX )ARMCPU *cpu =ARM_CPU (cs);CPUARMState *env = &cpu->env;@@ -425,6 +426,9 @@ int aarch64_gdb_set_tag_ctl_reg(CPUState *cs,
uint8_t*buf, int reg)Quoted 9 lines
target/arm/gdbstub64.c +3
-0arm_set_mte_tcf0 (env, tcf);return 1 ; +#else +return 0 ; +#endif }static void handle_q_memtag (GArray *params,void *user_ctx)
Re: [PATCH 13/14] bsd-user: Make compile for non-linux user-mode stuff
In reply to Warner LoshQuoted 7 lines · Warner Losh
We include the files that define
PR_MTE_TCF_SHIFTonly on Linux, but use them unconditionally. Restrict its use to Linux-only.Signed-off-by Warner Losh<[email protected]>--- target/arm/gdbstub64.c | 4 ++++ 1 file changed, 4 insertions(+)
r~
PATCH 14/14 bsd-user: Add aarch64 build to tree
In reply to Warner LoshAdd the aarch64 bsd-user fragments needed to build the new aarch64 code.
--- configs/targets/aarch64-bsd-user.mak | 3 +++ 1 file changed, 3 insertions(+) create mode 100644 configs/targets/aarch64-bsd-user.mak
configs/targets/aarch64-bsd-user.mak new file +3 -0
@@ -0,0 +1,3 @@ +TARGET_ARCH=aarch64 +TARGET_BASE_ARCH=arm +TARGET_XML_FILES= gdb-xml/aarch64-core.xml gdb-xml/aarch64-fpu.xml gdb-xml/aarch64-pauth.xml
Signature · 1 line
2.45.1
Re: [PATCH 14/14] bsd-user: Add aarch64 build to tree
Hi Warner,
Quoted 7 lines · Warner Losh
Add the aarch64 bsd-user fragments needed to build the new aarch64 code.
Signed-off-by Warner Losh <[email protected]>--- configs/targets/aarch64-bsd-user.mak | 3 +++ 1 file changed, 3 insertions(+)create mode 100644 configs/targets/aarch64-bsd-user.mak
Can we build aarch64 on Cirrus-CI? (not clear on
https://cirrus-ci.org/guide/FreeBSD/). If so, could you add a follow-up patch to build that on our CI, patching .gitlab-ci.d/cirrus.yml?
Regards,
Phil.
Re: [PATCH 14/14] bsd-user: Add aarch64 build to tree
Quoted 11 lines · Philippe Mathieu-Daudé
Hi Warner,
Quoted 7 lines · Warner Losh
Add the aarch64 bsd-user fragments needed to build the new aarch64 code.
Signed-off-by Warner Losh <[email protected]>--- configs/targets/aarch64-bsd-user.mak | 3 +++ 1 file changed, 3 insertions(+)create mode 100644 configs/targets/aarch64-bsd-user.mak
Quoted 5 lines
Can we build aarch64 on Cirrus-CI? (not clear on
https://cirrus-ci.org/guide/FreeBSD/). If so, could you add a follow-up patch to build that on our CI, patching.gitlab-ci.d/cirrus.yml?
We can build aarch64 host for bsd-user for sure. I'll see if we can do it in cirrus CI.
If so, I'll try to add it to cirrus.yml.
This patch series adds aarch64 guest...
Warner
Re: [PATCH 14/14] bsd-user: Add aarch64 build to tree
Quoted 23 lines · Warner Losh
On Mon, Jul 22, 2024 at 3:54 PM Philippe Mathieu-Daudé <[email protected] <mailto:[email protected]>> wrote:
Hi Warner,
On 22/7/24 23:43, Warner Losh wrote:
> Add the aarch64 bsd-user fragments needed to build the new
aarch64 code.
>
> Signed-off-by: Warner Losh <[email protected] <mailto:[email protected]>> > ---
> configs/targets/aarch64-bsd-user.mak | 3 +++
> 1 file changed, 3 insertions(+)
> create mode 100644 configs/targets/aarch64-bsd-user.makCan we build aarch64 on Cirrus-CI? (not clear on
https://cirrus-ci.org/guide/FreeBSD/
<https://cirrus-ci.org/guide/FreeBSD/>). If so, could you add a follow-up patch to build that on our CI, patching.gitlab-ci.d/cirrus.yml?
Quoted 4 lines
We can build aarch64 host for bsd-user for sure. I'll see if we can do it in cirrus CI. If so, I'll try to add it to cirrus.yml.
This patch series adds aarch64 guest...
Yes, we want to use a aarch64 FreeBSD host to build your FreeBSD aarch64 bsd-user guest and test it. Am I wrong?
Re: [PATCH 14/14] bsd-user: Add aarch64 build to tree
Quoted 24 lines · Philippe Mathieu-Daudé
Quoted 23 lines · Warner Losh
On Mon, Jul 22, 2024 at 3:54 PM Philippe Mathieu-Daudé <[email protected] <mailto:[email protected]>> wrote:
Hi Warner,
On 22/7/24 23:43, Warner Losh wrote:
> Add the aarch64 bsd-user fragments needed to build the new aarch64 code.
>
> Signed-off-by: Warner Losh <[email protected] <mailto:[email protected]>> > ---
> configs/targets/aarch64-bsd-user.mak | 3 +++
> 1 file changed, 3 insertions(+)
> create mode 100644 configs/targets/aarch64-bsd-user.makCan we build aarch64 on Cirrus-CI? (not clear on
https://cirrus-ci.org/guide/FreeBSD/
<https://cirrus-ci.org/guide/FreeBSD/>). If so, could you add a follow-up patch to build that on our CI, patching.gitlab-ci.d/cirrus.yml?
Quoted 8 lines
Quoted 4 lines
We can build aarch64 host for bsd-user for sure. I'll see if we can do it in cirrus CI. If so, I'll try to add it to cirrus.yml.
This patch series adds aarch64 guest...
Yes, we want to use a aarch64 FreeBSD host to build your FreeBSD aarch64 bsd-user guest and test it. Am I wrong?
This is adding guest support, so your suggestion is orthogonal.
r~
Re: [PATCH 14/14] bsd-user: Add aarch64 build to tree
Quoted 32 lines · Richard Henderson
Quoted 4 lines · Philippe Mathieu-Daudé
Quoted 3 lines · Warner Losh
On Mon, Jul 22, 2024 at 3:54 PM Philippe Mathieu-Daudé <
Quoted 9 lines
Quoted 9 lines
<mailto:[email protected]>> wrote:
Hi Warner,
On 22/7/24 23:43, Warner Losh wrote:
> Add the aarch64 bsd-user fragments needed to build the new
aarch64 code.
>
> Signed-off-by: Warner Losh <[email protected] <mailto:Quoted 5 lines
Quoted 5 lines
> ---
> configs/targets/aarch64-bsd-user.mak | 3 +++
> 1 file changed, 3 insertions(+)
> create mode 100644 configs/targets/aarch64-bsd-user.makQuoted 8 lines
Quoted 8 lines
Can we build aarch64 on Cirrus-CI? (not clear on
https://cirrus-ci.org/guide/FreeBSD/
<https://cirrus-ci.org/guide/FreeBSD/>). If so, could you add a follow-up patch to build that on our CI, patching.gitlab-ci.d/cirrus.yml?We can build aarch64 host for bsd-user for sure. I'll see if we can do
it in cirrus CI.
Quoted 2 lines
Quoted 2 lines
If so, I'll try to add it to cirrus.yml.
Quoted 8 lines
Quoted 5 lines
Quoted 1 line
This patch series adds aarch64 guest...
Yes, we want to use a aarch64 FreeBSD host to build your FreeBSD aarch64 bsd-user guest and test it. Am I wrong?
This is adding guest support, so your suggestion is orthogonal.
Yea... It's a good suggestion, but not what I'm working on right now...
Warner
Re: [PATCH 14/14] bsd-user: Add aarch64 build to tree
Oh, also, can I get a reviewed-by?
Warner
Quoted 50 lines · Warner Losh
Quoted 32 lines · Richard Henderson
Quoted 4 lines · Philippe Mathieu-Daudé
Quoted 3 lines · Warner Losh
On Mon, Jul 22, 2024 at 3:54 PM Philippe Mathieu-Daudé <
Quoted 9 lines
Quoted 9 lines
<mailto:[email protected]>> wrote:
Hi Warner,
On 22/7/24 23:43, Warner Losh wrote:
> Add the aarch64 bsd-user fragments needed to build the new
aarch64 code.
>
> Signed-off-by: Warner Losh <[email protected] <mailto:Quoted 5 lines
Quoted 5 lines
> ---
> configs/targets/aarch64-bsd-user.mak | 3 +++
> 1 file changed, 3 insertions(+)
> create mode 100644 configs/targets/aarch64-bsd-user.makQuoted 8 lines
Quoted 8 lines
Can we build aarch64 on Cirrus-CI? (not clear on
https://cirrus-ci.org/guide/FreeBSD/
<https://cirrus-ci.org/guide/FreeBSD/>). If so, could you add a follow-up patch to build that on our CI, patching.gitlab-ci.d/cirrus.yml?We can build aarch64 host for bsd-user for sure. I'll see if we can do
it in cirrus CI.
Quoted 2 lines
Quoted 2 lines
If so, I'll try to add it to cirrus.yml.
Quoted 8 lines
Quoted 5 lines
Quoted 1 line
This patch series adds aarch64 guest...
Yes, we want to use a aarch64 FreeBSD host to build your FreeBSD aarch64 bsd-user guest and test it. Am I wrong?
This is adding guest support, so your suggestion is orthogonal.
Yea... It's a good suggestion, but not what I'm working on right now...
Warner
Re: [PATCH 14/14] bsd-user: Add aarch64 build to tree
In reply to Richard HendersonQuoted 28 lines · Richard Henderson
Quoted 27 lines · Philippe Mathieu-Daudé
Quoted 26 lines · Warner Losh
On Mon, Jul 22, 2024 at 3:54 PM Philippe Mathieu-Daudé <[email protected] <mailto:[email protected]>> wrote:
Hi Warner,
On 22/7/24 23:43, Warner Losh wrote:
> Add the aarch64 bsd-user fragments needed to build the new aarch64 code.
>
> Signed-off-by: Warner Losh <[email protected] <mailto:[email protected]>> > ---
> configs/targets/aarch64-bsd-user.mak | 3 +++
> 1 file changed, 3 insertions(+)
> create mode 100644 configs/targets/aarch64-bsd-user.makCan we build aarch64 on Cirrus-CI? (not clear on
https://cirrus-ci.org/guide/FreeBSD/
<https://cirrus-ci.org/guide/FreeBSD/>). If so, could you add a follow-up patch to build that on our CI, patching.gitlab-ci.d/cirrus.yml?We can build aarch64 host for bsd-user for sure. I'll see if we can do it in cirrus CI. If so, I'll try to add it to cirrus.yml.
Quoted 7 lines
Quoted 5 lines
Quoted 1 line
This patch series adds aarch64 guest...
Yes, we want to use a aarch64 FreeBSD host to build your FreeBSD aarch64 bsd-user guest and test it. Am I wrong?
This is adding guest support, so your suggestion is orthogonal.
Ah, got it I guess, this series adds support for aarch64 user-mode emulation on any FreeBSD kernels, so IIUC we can test it on the current FreeBSD x86 job we have. Currently the x64-freebsd-13-build job uses:
CONFIGURE_ARGS:
--target-list-exclude=arm-softmmu,i386-softmmu,microblaze-softmmu,mips64el-softmmu,mipsel-softmmu,mips-softmmu,ppc-softmmu,sh4eb-softmmu,xtensa-softmmu
TEST_TARGETS: check
So the aarch64-bsd-user target should be selected automatically, is that correct?
(got it, my request to test aarch64 FreeBSD *host* is orthogonal).
Re: [PATCH 14/14] bsd-user: Add aarch64 build to tree
Quoted 44 lines · Philippe Mathieu-Daudé
Quoted 31 lines · Richard Henderson
Quoted 30 lines · Philippe Mathieu-Daudé
Quoted 29 lines · Warner Losh
On Mon, Jul 22, 2024 at 3:54 PM Philippe Mathieu-Daudé <[email protected] <mailto:[email protected]>> wrote:
Hi Warner,
On 22/7/24 23:43, Warner Losh wrote:
> Add the aarch64 bsd-user fragments needed to build the new
aarch64 code.
>
> Signed-off-by: Warner Losh <[email protected]
<mailto:[email protected]>>
> ---
> configs/targets/aarch64-bsd-user.mak | 3 +++
> 1 file changed, 3 insertions(+)
> create mode 100644 configs/targets/aarch64-bsd-user.makCan we build aarch64 on Cirrus-CI? (not clear on
https://cirrus-ci.org/guide/FreeBSD/
<https://cirrus-ci.org/guide/FreeBSD/>). If so, could you add a follow-up patch to build that on our CI, patching.gitlab-ci.d/cirrus.yml?We can build aarch64 host for bsd-user for sure. I'll see if we can do it in cirrus CI.
If so, I'll try to add it to cirrus.yml.Quoted 7 lines
Quoted 5 lines
Quoted 1 line
This patch series adds aarch64 guest...
Yes, we want to use a aarch64 FreeBSD host to build your FreeBSD aarch64 bsd-user guest and test it. Am I wrong?
This is adding guest support, so your suggestion is orthogonal.
Ah, got it I guess, this series adds support for aarch64 user-mode emulation on any FreeBSD kernels, so IIUC we can test it on the current FreeBSD x86 job we have. Currently the x64-freebsd-13-build job uses:
Quoted 8 lines
CONFIGURE_ARGS:--target-list-exclude=arm-softmmu,i386-softmmu,microblaze-softmmu,mips64el-softmmu,mipsel-softmmu,mips-softmmu,ppc-softmmu,sh4eb-softmmu,xtensa-softmmu
TEST_TARGETS: checkSo the aarch64-bsd-user target should be selected automatically, is that correct?
Yes. It's automatically added to the build just by this file existing.
So it should at least be compile tested by everything.
Warner
Quoted 2 lines
(got it, my request to test aarch64 FreeBSD *host* is orthogonal).
Re: [PATCH 14/14] bsd-user: Add aarch64 build to tree
In reply to Warner LoshQuoted 7 lines · Warner Losh
Add the aarch64 bsd-user fragments needed to build the new aarch64 code.
Signed-off-by Warner Losh<[email protected]>--- configs/targets/aarch64-bsd-user.mak | 3 +++ 1 file changed, 3 insertions(+)create mode 100644 configs/targets/aarch64-bsd-user.mak
r~