qemu/target/i386/smm_helper.c

350 lines
13 KiB
C
Raw Normal View History

/*
* x86 SMM helpers
*
* Copyright (c) 2003 Fabrice Bellard
*
* 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"
tcg: drop global lock during TCG code execution This finally allows TCG to benefit from the iothread introduction: Drop the global mutex while running pure TCG CPU code. Reacquire the lock when entering MMIO or PIO emulation, or when leaving the TCG loop. We have to revert a few optimization for the current TCG threading model, namely kicking the TCG thread in qemu_mutex_lock_iothread and not kicking it in qemu_cpu_kick. We also need to disable RAM block reordering until we have a more efficient locking mechanism at hand. Still, a Linux x86 UP guest and my Musicpal ARM model boot fine here. These numbers demonstrate where we gain something: 20338 jan 20 0 331m 75m 6904 R 99 0.9 0:50.95 qemu-system-arm 20337 jan 20 0 331m 75m 6904 S 20 0.9 0:26.50 qemu-system-arm The guest CPU was fully loaded, but the iothread could still run mostly independent on a second core. Without the patch we don't get beyond 32206 jan 20 0 330m 73m 7036 R 82 0.9 1:06.00 qemu-system-arm 32204 jan 20 0 330m 73m 7036 S 21 0.9 0:17.03 qemu-system-arm We don't benefit significantly, though, when the guest is not fully loading a host CPU. Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> Message-Id: <1439220437-23957-10-git-send-email-fred.konrad@greensocs.com> [FK: Rebase, fix qemu_devices_reset deadlock, rm address_space_* mutex] Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com> [EGC: fixed iothread lock for cpu-exec IRQ handling] Signed-off-by: Emilio G. Cota <cota@braap.org> [AJB: -smp single-threaded fix, clean commit msg, BQL fixes] Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <rth@twiddle.net> Reviewed-by: Pranith Kumar <bobby.prani@gmail.com> [PM: target-arm changes] Acked-by: Peter Maydell <peter.maydell@linaro.org>
2017-02-23 21:29:11 +03:00
#include "qemu/main-loop.h"
#include "cpu.h"
#include "exec/helper-proto.h"
#include "exec/log.h"
/* SMM support */
#if defined(CONFIG_USER_ONLY)
void do_smm_enter(X86CPU *cpu)
{
}
void helper_rsm(CPUX86State *env)
{
}
#else
#ifdef TARGET_X86_64
#define SMM_REVISION_ID 0x00020064
#else
#define SMM_REVISION_ID 0x00020000
#endif
tcg: drop global lock during TCG code execution This finally allows TCG to benefit from the iothread introduction: Drop the global mutex while running pure TCG CPU code. Reacquire the lock when entering MMIO or PIO emulation, or when leaving the TCG loop. We have to revert a few optimization for the current TCG threading model, namely kicking the TCG thread in qemu_mutex_lock_iothread and not kicking it in qemu_cpu_kick. We also need to disable RAM block reordering until we have a more efficient locking mechanism at hand. Still, a Linux x86 UP guest and my Musicpal ARM model boot fine here. These numbers demonstrate where we gain something: 20338 jan 20 0 331m 75m 6904 R 99 0.9 0:50.95 qemu-system-arm 20337 jan 20 0 331m 75m 6904 S 20 0.9 0:26.50 qemu-system-arm The guest CPU was fully loaded, but the iothread could still run mostly independent on a second core. Without the patch we don't get beyond 32206 jan 20 0 330m 73m 7036 R 82 0.9 1:06.00 qemu-system-arm 32204 jan 20 0 330m 73m 7036 S 21 0.9 0:17.03 qemu-system-arm We don't benefit significantly, though, when the guest is not fully loading a host CPU. Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> Message-Id: <1439220437-23957-10-git-send-email-fred.konrad@greensocs.com> [FK: Rebase, fix qemu_devices_reset deadlock, rm address_space_* mutex] Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com> [EGC: fixed iothread lock for cpu-exec IRQ handling] Signed-off-by: Emilio G. Cota <cota@braap.org> [AJB: -smp single-threaded fix, clean commit msg, BQL fixes] Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <rth@twiddle.net> Reviewed-by: Pranith Kumar <bobby.prani@gmail.com> [PM: target-arm changes] Acked-by: Peter Maydell <peter.maydell@linaro.org>
2017-02-23 21:29:11 +03:00
/* Called with iothread lock taken */
void cpu_smm_update(X86CPU *cpu)
{
CPUX86State *env = &cpu->env;
bool smm_enabled = (env->hflags & HF_SMM_MASK);
tcg: drop global lock during TCG code execution This finally allows TCG to benefit from the iothread introduction: Drop the global mutex while running pure TCG CPU code. Reacquire the lock when entering MMIO or PIO emulation, or when leaving the TCG loop. We have to revert a few optimization for the current TCG threading model, namely kicking the TCG thread in qemu_mutex_lock_iothread and not kicking it in qemu_cpu_kick. We also need to disable RAM block reordering until we have a more efficient locking mechanism at hand. Still, a Linux x86 UP guest and my Musicpal ARM model boot fine here. These numbers demonstrate where we gain something: 20338 jan 20 0 331m 75m 6904 R 99 0.9 0:50.95 qemu-system-arm 20337 jan 20 0 331m 75m 6904 S 20 0.9 0:26.50 qemu-system-arm The guest CPU was fully loaded, but the iothread could still run mostly independent on a second core. Without the patch we don't get beyond 32206 jan 20 0 330m 73m 7036 R 82 0.9 1:06.00 qemu-system-arm 32204 jan 20 0 330m 73m 7036 S 21 0.9 0:17.03 qemu-system-arm We don't benefit significantly, though, when the guest is not fully loading a host CPU. Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> Message-Id: <1439220437-23957-10-git-send-email-fred.konrad@greensocs.com> [FK: Rebase, fix qemu_devices_reset deadlock, rm address_space_* mutex] Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com> [EGC: fixed iothread lock for cpu-exec IRQ handling] Signed-off-by: Emilio G. Cota <cota@braap.org> [AJB: -smp single-threaded fix, clean commit msg, BQL fixes] Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <rth@twiddle.net> Reviewed-by: Pranith Kumar <bobby.prani@gmail.com> [PM: target-arm changes] Acked-by: Peter Maydell <peter.maydell@linaro.org>
2017-02-23 21:29:11 +03:00
g_assert(qemu_mutex_iothread_locked());
if (cpu->smram) {
memory_region_set_enabled(cpu->smram, smm_enabled);
}
}
void do_smm_enter(X86CPU *cpu)
{
CPUX86State *env = &cpu->env;
CPUState *cs = CPU(cpu);
target_ulong sm_state;
SegmentCache *dt;
int i, offset;
qemu_log_mask(CPU_LOG_INT, "SMM: enter\n");
log_cpu_state_mask(CPU_LOG_INT, CPU(cpu), CPU_DUMP_CCOP);
env->hflags |= HF_SMM_MASK;
if (env->hflags2 & HF2_NMI_MASK) {
env->hflags2 |= HF2_SMM_INSIDE_NMI_MASK;
} else {
env->hflags2 |= HF2_NMI_MASK;
}
cpu_smm_update(cpu);
sm_state = env->smbase + 0x8000;
#ifdef TARGET_X86_64
for (i = 0; i < 6; i++) {
dt = &env->segs[i];
offset = 0x7e00 + i * 16;
x86_stw_phys(cs, sm_state + offset, dt->selector);
x86_stw_phys(cs, sm_state + offset + 2, (dt->flags >> 8) & 0xf0ff);
x86_stl_phys(cs, sm_state + offset + 4, dt->limit);
x86_stq_phys(cs, sm_state + offset + 8, dt->base);
}
x86_stq_phys(cs, sm_state + 0x7e68, env->gdt.base);
x86_stl_phys(cs, sm_state + 0x7e64, env->gdt.limit);
x86_stw_phys(cs, sm_state + 0x7e70, env->ldt.selector);
x86_stq_phys(cs, sm_state + 0x7e78, env->ldt.base);
x86_stl_phys(cs, sm_state + 0x7e74, env->ldt.limit);
x86_stw_phys(cs, sm_state + 0x7e72, (env->ldt.flags >> 8) & 0xf0ff);
x86_stq_phys(cs, sm_state + 0x7e88, env->idt.base);
x86_stl_phys(cs, sm_state + 0x7e84, env->idt.limit);
x86_stw_phys(cs, sm_state + 0x7e90, env->tr.selector);
x86_stq_phys(cs, sm_state + 0x7e98, env->tr.base);
x86_stl_phys(cs, sm_state + 0x7e94, env->tr.limit);
x86_stw_phys(cs, sm_state + 0x7e92, (env->tr.flags >> 8) & 0xf0ff);
/* ??? Vol 1, 16.5.6 Intel MPX and SMM says that IA32_BNDCFGS
is saved at offset 7ED0. Vol 3, 34.4.1.1, Table 32-2, has
7EA0-7ED7 as "reserved". What's this, and what's really
supposed to happen? */
x86_stq_phys(cs, sm_state + 0x7ed0, env->efer);
x86_stq_phys(cs, sm_state + 0x7ff8, env->regs[R_EAX]);
x86_stq_phys(cs, sm_state + 0x7ff0, env->regs[R_ECX]);
x86_stq_phys(cs, sm_state + 0x7fe8, env->regs[R_EDX]);
x86_stq_phys(cs, sm_state + 0x7fe0, env->regs[R_EBX]);
x86_stq_phys(cs, sm_state + 0x7fd8, env->regs[R_ESP]);
x86_stq_phys(cs, sm_state + 0x7fd0, env->regs[R_EBP]);
x86_stq_phys(cs, sm_state + 0x7fc8, env->regs[R_ESI]);
x86_stq_phys(cs, sm_state + 0x7fc0, env->regs[R_EDI]);
for (i = 8; i < 16; i++) {
x86_stq_phys(cs, sm_state + 0x7ff8 - i * 8, env->regs[i]);
}
x86_stq_phys(cs, sm_state + 0x7f78, env->eip);
x86_stl_phys(cs, sm_state + 0x7f70, cpu_compute_eflags(env));
x86_stl_phys(cs, sm_state + 0x7f68, env->dr[6]);
x86_stl_phys(cs, sm_state + 0x7f60, env->dr[7]);
x86_stl_phys(cs, sm_state + 0x7f48, env->cr[4]);
x86_stq_phys(cs, sm_state + 0x7f50, env->cr[3]);
x86_stl_phys(cs, sm_state + 0x7f58, env->cr[0]);
x86_stl_phys(cs, sm_state + 0x7efc, SMM_REVISION_ID);
x86_stl_phys(cs, sm_state + 0x7f00, env->smbase);
#else
x86_stl_phys(cs, sm_state + 0x7ffc, env->cr[0]);
x86_stl_phys(cs, sm_state + 0x7ff8, env->cr[3]);
x86_stl_phys(cs, sm_state + 0x7ff4, cpu_compute_eflags(env));
x86_stl_phys(cs, sm_state + 0x7ff0, env->eip);
x86_stl_phys(cs, sm_state + 0x7fec, env->regs[R_EDI]);
x86_stl_phys(cs, sm_state + 0x7fe8, env->regs[R_ESI]);
x86_stl_phys(cs, sm_state + 0x7fe4, env->regs[R_EBP]);
x86_stl_phys(cs, sm_state + 0x7fe0, env->regs[R_ESP]);
x86_stl_phys(cs, sm_state + 0x7fdc, env->regs[R_EBX]);
x86_stl_phys(cs, sm_state + 0x7fd8, env->regs[R_EDX]);
x86_stl_phys(cs, sm_state + 0x7fd4, env->regs[R_ECX]);
x86_stl_phys(cs, sm_state + 0x7fd0, env->regs[R_EAX]);
x86_stl_phys(cs, sm_state + 0x7fcc, env->dr[6]);
x86_stl_phys(cs, sm_state + 0x7fc8, env->dr[7]);
x86_stl_phys(cs, sm_state + 0x7fc4, env->tr.selector);
x86_stl_phys(cs, sm_state + 0x7f64, env->tr.base);
x86_stl_phys(cs, sm_state + 0x7f60, env->tr.limit);
x86_stl_phys(cs, sm_state + 0x7f5c, (env->tr.flags >> 8) & 0xf0ff);
x86_stl_phys(cs, sm_state + 0x7fc0, env->ldt.selector);
x86_stl_phys(cs, sm_state + 0x7f80, env->ldt.base);
x86_stl_phys(cs, sm_state + 0x7f7c, env->ldt.limit);
x86_stl_phys(cs, sm_state + 0x7f78, (env->ldt.flags >> 8) & 0xf0ff);
x86_stl_phys(cs, sm_state + 0x7f74, env->gdt.base);
x86_stl_phys(cs, sm_state + 0x7f70, env->gdt.limit);
x86_stl_phys(cs, sm_state + 0x7f58, env->idt.base);
x86_stl_phys(cs, sm_state + 0x7f54, env->idt.limit);
for (i = 0; i < 6; i++) {
dt = &env->segs[i];
if (i < 3) {
offset = 0x7f84 + i * 12;
} else {
offset = 0x7f2c + (i - 3) * 12;
}
x86_stl_phys(cs, sm_state + 0x7fa8 + i * 4, dt->selector);
x86_stl_phys(cs, sm_state + offset + 8, dt->base);
x86_stl_phys(cs, sm_state + offset + 4, dt->limit);
x86_stl_phys(cs, sm_state + offset, (dt->flags >> 8) & 0xf0ff);
}
x86_stl_phys(cs, sm_state + 0x7f14, env->cr[4]);
x86_stl_phys(cs, sm_state + 0x7efc, SMM_REVISION_ID);
x86_stl_phys(cs, sm_state + 0x7ef8, env->smbase);
#endif
/* init SMM cpu state */
#ifdef TARGET_X86_64
cpu_load_efer(env, 0);
#endif
cpu_load_eflags(env, 0, ~(CC_O | CC_S | CC_Z | CC_A | CC_P | CC_C |
DF_MASK));
env->eip = 0x00008000;
cpu_x86_update_cr0(env,
env->cr[0] & ~(CR0_PE_MASK | CR0_EM_MASK | CR0_TS_MASK |
CR0_PG_MASK));
cpu_x86_update_cr4(env, 0);
env->dr[7] = 0x00000400;
cpu_x86_load_seg_cache(env, R_CS, (env->smbase >> 4) & 0xffff, env->smbase,
0xffffffff,
DESC_P_MASK | DESC_S_MASK | DESC_W_MASK |
DESC_G_MASK | DESC_A_MASK);
cpu_x86_load_seg_cache(env, R_DS, 0, 0, 0xffffffff,
DESC_P_MASK | DESC_S_MASK | DESC_W_MASK |
DESC_G_MASK | DESC_A_MASK);
cpu_x86_load_seg_cache(env, R_ES, 0, 0, 0xffffffff,
DESC_P_MASK | DESC_S_MASK | DESC_W_MASK |
DESC_G_MASK | DESC_A_MASK);
cpu_x86_load_seg_cache(env, R_SS, 0, 0, 0xffffffff,
DESC_P_MASK | DESC_S_MASK | DESC_W_MASK |
DESC_G_MASK | DESC_A_MASK);
cpu_x86_load_seg_cache(env, R_FS, 0, 0, 0xffffffff,
DESC_P_MASK | DESC_S_MASK | DESC_W_MASK |
DESC_G_MASK | DESC_A_MASK);
cpu_x86_load_seg_cache(env, R_GS, 0, 0, 0xffffffff,
DESC_P_MASK | DESC_S_MASK | DESC_W_MASK |
DESC_G_MASK | DESC_A_MASK);
}
void helper_rsm(CPUX86State *env)
{
X86CPU *cpu = x86_env_get_cpu(env);
CPUState *cs = CPU(cpu);
target_ulong sm_state;
int i, offset;
uint32_t val;
sm_state = env->smbase + 0x8000;
#ifdef TARGET_X86_64
cpu_load_efer(env, x86_ldq_phys(cs, sm_state + 0x7ed0));
env->gdt.base = x86_ldq_phys(cs, sm_state + 0x7e68);
env->gdt.limit = x86_ldl_phys(cs, sm_state + 0x7e64);
env->ldt.selector = x86_lduw_phys(cs, sm_state + 0x7e70);
env->ldt.base = x86_ldq_phys(cs, sm_state + 0x7e78);
env->ldt.limit = x86_ldl_phys(cs, sm_state + 0x7e74);
env->ldt.flags = (x86_lduw_phys(cs, sm_state + 0x7e72) & 0xf0ff) << 8;
env->idt.base = x86_ldq_phys(cs, sm_state + 0x7e88);
env->idt.limit = x86_ldl_phys(cs, sm_state + 0x7e84);
env->tr.selector = x86_lduw_phys(cs, sm_state + 0x7e90);
env->tr.base = x86_ldq_phys(cs, sm_state + 0x7e98);
env->tr.limit = x86_ldl_phys(cs, sm_state + 0x7e94);
env->tr.flags = (x86_lduw_phys(cs, sm_state + 0x7e92) & 0xf0ff) << 8;
env->regs[R_EAX] = x86_ldq_phys(cs, sm_state + 0x7ff8);
env->regs[R_ECX] = x86_ldq_phys(cs, sm_state + 0x7ff0);
env->regs[R_EDX] = x86_ldq_phys(cs, sm_state + 0x7fe8);
env->regs[R_EBX] = x86_ldq_phys(cs, sm_state + 0x7fe0);
env->regs[R_ESP] = x86_ldq_phys(cs, sm_state + 0x7fd8);
env->regs[R_EBP] = x86_ldq_phys(cs, sm_state + 0x7fd0);
env->regs[R_ESI] = x86_ldq_phys(cs, sm_state + 0x7fc8);
env->regs[R_EDI] = x86_ldq_phys(cs, sm_state + 0x7fc0);
for (i = 8; i < 16; i++) {
env->regs[i] = x86_ldq_phys(cs, sm_state + 0x7ff8 - i * 8);
}
env->eip = x86_ldq_phys(cs, sm_state + 0x7f78);
cpu_load_eflags(env, x86_ldl_phys(cs, sm_state + 0x7f70),
~(CC_O | CC_S | CC_Z | CC_A | CC_P | CC_C | DF_MASK));
env->dr[6] = x86_ldl_phys(cs, sm_state + 0x7f68);
env->dr[7] = x86_ldl_phys(cs, sm_state + 0x7f60);
cpu_x86_update_cr4(env, x86_ldl_phys(cs, sm_state + 0x7f48));
cpu_x86_update_cr3(env, x86_ldq_phys(cs, sm_state + 0x7f50));
cpu_x86_update_cr0(env, x86_ldl_phys(cs, sm_state + 0x7f58));
for (i = 0; i < 6; i++) {
offset = 0x7e00 + i * 16;
cpu_x86_load_seg_cache(env, i,
x86_lduw_phys(cs, sm_state + offset),
x86_ldq_phys(cs, sm_state + offset + 8),
x86_ldl_phys(cs, sm_state + offset + 4),
(x86_lduw_phys(cs, sm_state + offset + 2) &
0xf0ff) << 8);
}
val = x86_ldl_phys(cs, sm_state + 0x7efc); /* revision ID */
if (val & 0x20000) {
env->smbase = x86_ldl_phys(cs, sm_state + 0x7f00);
}
#else
cpu_x86_update_cr0(env, x86_ldl_phys(cs, sm_state + 0x7ffc));
cpu_x86_update_cr3(env, x86_ldl_phys(cs, sm_state + 0x7ff8));
cpu_load_eflags(env, x86_ldl_phys(cs, sm_state + 0x7ff4),
~(CC_O | CC_S | CC_Z | CC_A | CC_P | CC_C | DF_MASK));
env->eip = x86_ldl_phys(cs, sm_state + 0x7ff0);
env->regs[R_EDI] = x86_ldl_phys(cs, sm_state + 0x7fec);
env->regs[R_ESI] = x86_ldl_phys(cs, sm_state + 0x7fe8);
env->regs[R_EBP] = x86_ldl_phys(cs, sm_state + 0x7fe4);
env->regs[R_ESP] = x86_ldl_phys(cs, sm_state + 0x7fe0);
env->regs[R_EBX] = x86_ldl_phys(cs, sm_state + 0x7fdc);
env->regs[R_EDX] = x86_ldl_phys(cs, sm_state + 0x7fd8);
env->regs[R_ECX] = x86_ldl_phys(cs, sm_state + 0x7fd4);
env->regs[R_EAX] = x86_ldl_phys(cs, sm_state + 0x7fd0);
env->dr[6] = x86_ldl_phys(cs, sm_state + 0x7fcc);
env->dr[7] = x86_ldl_phys(cs, sm_state + 0x7fc8);
env->tr.selector = x86_ldl_phys(cs, sm_state + 0x7fc4) & 0xffff;
env->tr.base = x86_ldl_phys(cs, sm_state + 0x7f64);
env->tr.limit = x86_ldl_phys(cs, sm_state + 0x7f60);
env->tr.flags = (x86_ldl_phys(cs, sm_state + 0x7f5c) & 0xf0ff) << 8;
env->ldt.selector = x86_ldl_phys(cs, sm_state + 0x7fc0) & 0xffff;
env->ldt.base = x86_ldl_phys(cs, sm_state + 0x7f80);
env->ldt.limit = x86_ldl_phys(cs, sm_state + 0x7f7c);
env->ldt.flags = (x86_ldl_phys(cs, sm_state + 0x7f78) & 0xf0ff) << 8;
env->gdt.base = x86_ldl_phys(cs, sm_state + 0x7f74);
env->gdt.limit = x86_ldl_phys(cs, sm_state + 0x7f70);
env->idt.base = x86_ldl_phys(cs, sm_state + 0x7f58);
env->idt.limit = x86_ldl_phys(cs, sm_state + 0x7f54);
for (i = 0; i < 6; i++) {
if (i < 3) {
offset = 0x7f84 + i * 12;
} else {
offset = 0x7f2c + (i - 3) * 12;
}
cpu_x86_load_seg_cache(env, i,
x86_ldl_phys(cs,
sm_state + 0x7fa8 + i * 4) & 0xffff,
x86_ldl_phys(cs, sm_state + offset + 8),
x86_ldl_phys(cs, sm_state + offset + 4),
(x86_ldl_phys(cs,
sm_state + offset) & 0xf0ff) << 8);
}
cpu_x86_update_cr4(env, x86_ldl_phys(cs, sm_state + 0x7f14));
val = x86_ldl_phys(cs, sm_state + 0x7efc); /* revision ID */
if (val & 0x20000) {
env->smbase = x86_ldl_phys(cs, sm_state + 0x7ef8);
}
#endif
if ((env->hflags2 & HF2_SMM_INSIDE_NMI_MASK) == 0) {
env->hflags2 &= ~HF2_NMI_MASK;
}
env->hflags2 &= ~HF2_SMM_INSIDE_NMI_MASK;
env->hflags &= ~HF_SMM_MASK;
tcg: drop global lock during TCG code execution This finally allows TCG to benefit from the iothread introduction: Drop the global mutex while running pure TCG CPU code. Reacquire the lock when entering MMIO or PIO emulation, or when leaving the TCG loop. We have to revert a few optimization for the current TCG threading model, namely kicking the TCG thread in qemu_mutex_lock_iothread and not kicking it in qemu_cpu_kick. We also need to disable RAM block reordering until we have a more efficient locking mechanism at hand. Still, a Linux x86 UP guest and my Musicpal ARM model boot fine here. These numbers demonstrate where we gain something: 20338 jan 20 0 331m 75m 6904 R 99 0.9 0:50.95 qemu-system-arm 20337 jan 20 0 331m 75m 6904 S 20 0.9 0:26.50 qemu-system-arm The guest CPU was fully loaded, but the iothread could still run mostly independent on a second core. Without the patch we don't get beyond 32206 jan 20 0 330m 73m 7036 R 82 0.9 1:06.00 qemu-system-arm 32204 jan 20 0 330m 73m 7036 S 21 0.9 0:17.03 qemu-system-arm We don't benefit significantly, though, when the guest is not fully loading a host CPU. Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> Message-Id: <1439220437-23957-10-git-send-email-fred.konrad@greensocs.com> [FK: Rebase, fix qemu_devices_reset deadlock, rm address_space_* mutex] Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com> [EGC: fixed iothread lock for cpu-exec IRQ handling] Signed-off-by: Emilio G. Cota <cota@braap.org> [AJB: -smp single-threaded fix, clean commit msg, BQL fixes] Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <rth@twiddle.net> Reviewed-by: Pranith Kumar <bobby.prani@gmail.com> [PM: target-arm changes] Acked-by: Peter Maydell <peter.maydell@linaro.org>
2017-02-23 21:29:11 +03:00
qemu_mutex_lock_iothread();
cpu_smm_update(cpu);
tcg: drop global lock during TCG code execution This finally allows TCG to benefit from the iothread introduction: Drop the global mutex while running pure TCG CPU code. Reacquire the lock when entering MMIO or PIO emulation, or when leaving the TCG loop. We have to revert a few optimization for the current TCG threading model, namely kicking the TCG thread in qemu_mutex_lock_iothread and not kicking it in qemu_cpu_kick. We also need to disable RAM block reordering until we have a more efficient locking mechanism at hand. Still, a Linux x86 UP guest and my Musicpal ARM model boot fine here. These numbers demonstrate where we gain something: 20338 jan 20 0 331m 75m 6904 R 99 0.9 0:50.95 qemu-system-arm 20337 jan 20 0 331m 75m 6904 S 20 0.9 0:26.50 qemu-system-arm The guest CPU was fully loaded, but the iothread could still run mostly independent on a second core. Without the patch we don't get beyond 32206 jan 20 0 330m 73m 7036 R 82 0.9 1:06.00 qemu-system-arm 32204 jan 20 0 330m 73m 7036 S 21 0.9 0:17.03 qemu-system-arm We don't benefit significantly, though, when the guest is not fully loading a host CPU. Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> Message-Id: <1439220437-23957-10-git-send-email-fred.konrad@greensocs.com> [FK: Rebase, fix qemu_devices_reset deadlock, rm address_space_* mutex] Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com> [EGC: fixed iothread lock for cpu-exec IRQ handling] Signed-off-by: Emilio G. Cota <cota@braap.org> [AJB: -smp single-threaded fix, clean commit msg, BQL fixes] Signed-off-by: Alex Bennée <alex.bennee@linaro.org> Reviewed-by: Richard Henderson <rth@twiddle.net> Reviewed-by: Pranith Kumar <bobby.prani@gmail.com> [PM: target-arm changes] Acked-by: Peter Maydell <peter.maydell@linaro.org>
2017-02-23 21:29:11 +03:00
qemu_mutex_unlock_iothread();
qemu_log_mask(CPU_LOG_INT, "SMM: after RSM\n");
log_cpu_state_mask(CPU_LOG_INT, CPU(cpu), CPU_DUMP_CCOP);
}
#endif /* !CONFIG_USER_ONLY */