Guest debugging support for KVM
PATCH v3 1/2 Guest debugging support for KVM
This is a backport of the guest debugging support for the KVM accelerator that is now part of the KVM tree. It implements the reworked KVM kernel API for guest debugging (KVM_CAP_SET_GUEST_DEBUG) which is not yet part of any mainline kernel but will probably be 2.6.30 stuff. So far supported is x86, but PPC is expected to catch up soon.
Core features are:
- unlimited soft-breakpoints via code patching
- hardware-assisted x86 breakpoints and watchpoints
Changes in this version:
- use generic hook cpu_synchronize_state to transfer registers between user space and kvm
- push kvm_sw_breakpoints into KVMState
---
exec.c | 10 ++- gdbstub.c | 24 +++++-- gdbstub.h | 7 ++ kvm-all.c | 173 ++++++++++++++++++++++++++++++++++++++++++++++++ kvm.h | 51 ++++++++++++++ target-i386/kvm.c | 191 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 6 files changed, 446 insertions(+), 10 deletions(-)
exec.c +7 -3
@@ -1457,9 +1457,13 @@ void cpu_single_step(CPUState *env, int enabled)#if defined(TARGET_HAS_ICE )if (env->singlestep_enabled != enabled) { env->singlestep_enabled = enabled; -/* must flush all the translated code to avoid inconsistancies */ -/* XXX: only flush what is necessary */ -tb_flush (env); +if (kvm_enabled ()) +kvm_update_guest_debug (env,0 ); +else { +/* must flush all the translated code to avoid inconsistancies */ +/* XXX: only flush what is necessary */ +tb_flush (env); + } }#endif }
gdbstub.c +17 -7
@@ -39,6 +39,7 @@#define MAX_PACKET_LENGTH 4096#include "qemu_socket.h" +#include "kvm.h" enum {
@@ -1418,13 +1419,6 @@ void gdb_register_coprocessor(CPUState * env, } } -/* GDB breakpoint/watchpoint types */ -#define GDB_BREAKPOINT_SW 0 -#define GDB_BREAKPOINT_HW 1 -#define GDB_WATCHPOINT_WRITE 2 -#define GDB_WATCHPOINT_READ 3 -#define GDB_WATCHPOINT_ACCESS 4 -#ifndef CONFIG_USER_ONLY static const int xlat_gdb_type[] = { [GDB_WATCHPOINT_WRITE] = BP_GDB | BP_MEM_WRITE,
@@ -1438,6 +1432,9 @@ static int gdb_breakpoint_insert(target_ulong addr, target_ulong len, int type)CPUState *env;int err =0 ; +if (kvm_enabled ()) +return kvm_insert_breakpoint (gdbserver_state->c_cpu, addr, len, type); +switch (type) {case GDB_BREAKPOINT_SW :case GDB_BREAKPOINT_HW :
@@ -1469,6 +1466,9 @@ static int gdb_breakpoint_remove(target_ulong addr, target_ulong len, int type)CPUState *env;int err =0 ; +if (kvm_enabled ()) +return kvm_remove_breakpoint (gdbserver_state->c_cpu, addr, len, type); +switch (type) {case GDB_BREAKPOINT_SW :case GDB_BREAKPOINT_HW :
@@ -1498,6 +1498,11 @@ static void gdb_breakpoint_remove_all(void) {CPUState *env; +if (kvm_enabled ()) { +kvm_remove_all_breakpoints (gdbserver_state->c_cpu); +return ; + } +for (env = first_cpu; env !=NULL ; env = env->next_cpu) {cpu_breakpoint_remove_all (env,BP_GDB );#ifndef CONFIG_USER_ONLY
@@ -1538,6 +1543,7 @@ static int gdb_handle_packet(GDBState *s, const char *line_buf) addr =strtoull (p, (char **)&p,16 );#if defined(TARGET_I386 ) s->c_cpu->eip = addr; +cpu_synchronize_state (s->c_cpu,1 );#elif defined (TARGET_PPC ) s->c_cpu->nip = addr;#elif defined (TARGET_SPARC )
@@ -1579,6 +1585,7 @@ static int gdb_handle_packet(GDBState *s, const char *line_buf) addr =strtoull (p, (char **)&p,16 );#if defined(TARGET_I386 ) s->c_cpu->eip = addr; +cpu_synchronize_state (s->c_cpu,1 );#elif defined (TARGET_PPC ) s->c_cpu->nip = addr;#elif defined (TARGET_SPARC )
@@ -1624,6 +1631,7 @@ static int gdb_handle_packet(GDBState *s, const char *line_buf) }break ;case 'g' : +cpu_synchronize_state (s->g_cpu,0 ); len =0 ;for (addr =0 ; addr < num_g_regs; addr++) { reg_size =gdb_read_register (s->g_cpu, mem_buf + len, addr);
@@ -1641,6 +1649,7 @@ static int gdb_handle_packet(GDBState *s, const char *line_buf) len -= reg_size; registers += reg_size; } +cpu_synchronize_state (s->g_cpu,1 );put_packet (s,"OK" );break ;case 'm' :
@@ -1799,6 +1808,7 @@ static int gdb_handle_packet(GDBState *s, const char *line_buf) thread =strtoull (p+16 , (char **)&p,16 );for (env = first_cpu; env !=NULL ; env = env->next_cpu)if (env->cpu_index +1 == thread) { +cpu_synchronize_state (env,0 ); len =snprintf ((char *)mem_buf,sizeof (mem_buf),"CPU#%d [%s]" , env->cpu_index, env->halted ?"halted " :"running" );
gdbstub.h +7 -0
@@ -3,6 +3,13 @@#define DEFAULT_GDBSTUB_PORT "1234" +/* GDB breakpoint/watchpoint types */ +#define GDB_BREAKPOINT_SW 0 +#define GDB_BREAKPOINT_HW 1 +#define GDB_WATCHPOINT_WRITE 2 +#define GDB_WATCHPOINT_READ 3 +#define GDB_WATCHPOINT_ACCESS 4 +typedef void (*gdb_syscall_complete_cb )(CPUState *env,target_ulong ret,target_ulong err);
kvm-all.c +173 -0
200 lines not on this page · raw message
kvm.h +51 -0
67 lines not on this page · raw message
target-i386/kvm.c +191 -0
200 lines not on this page · raw message
Quoted 9 lines · Jan Kiszka
This is a backport of the guest debugging support for the KVM
accelerator that is now part of the KVM tree. It implements the reworked KVM kernel API for guest debugging (KVM_CAP_SET_GUEST_DEBUG) which is not yet part of any mainline kernel but will probably be 2.6.30 stuff. So far supported is x86, but PPC is expected to catch up soon.Core features are:
- unlimited soft-breakpoints via code patching
- hardware-assisted x86 breakpoints and watchpoints
The statically dimensioned arrays (hw_breakpoint[4], single hw_watchpoint) make me wonder what would happen in the SMP case. Or is only one CPU ever targeted at a time for debugging?
Quoted 7 lines · Blue Swirl
Quoted 6 lines · Jan Kiszka
This is a backport of the guest debugging support for the KVM
accelerator that is now part of the KVM tree. It implements the reworked KVM kernel API for guest debugging (KVM_CAP_SET_GUEST_DEBUG) which is not yet part of any mainline kernel but will probably be 2.6.30 stuff. So far supported is x86, but PPC is expected to catch up soon.
Quoted 8 lines
Quoted 3 lines
Core features are:
- unlimited soft-breakpoints via code patching
- hardware-assisted x86 breakpoints and watchpointsThe statically dimensioned arrays (hw_breakpoint[4], single
hw_watchpoint) make me wonder what would happen in the SMP case. Or is only one CPU ever targeted at a time for debugging?
We have 4 hardware watch- or breakpoint on x86. KVM is able to set them per VCPU, that's why all relevant functions take a CPUState as argument. So that API is highly flexible.
But, and that's the "famous" :) topic discussed here before /wrt plain
qemu: gdb has no multicore or SMP model yet so that per-CPU breakpoints could be assigned and properly managed. Therefore qemu applies break- and watchpoints to all virtual CPUs, and with this patch also in kvm mode. That covers most of today's use cases, and with some care you can even apply it to systems which run different memory mappings and different code on their CPUs.
[ Reminds me that I should repost my vCont patch... ]
Jan
Signature · 2 lines
Siemens AG, Corporate Technology, CT SE 2 Corporate Competence Center Embedded Linux
Anthony Liguori
AppliedQuoted 10 lines · Jan Kiszka
This is a backport of the guest debugging support for the KVM accelerator that is now part of the KVM tree. It implements the reworked KVM kernel API for guest debugging (
KVM_CAP_SET_GUEST_DEBUG) which is not yet part of any mainline kernel but will probably be 2.6.30 stuff. So far supported is x86, but PPC is expected to catch up soon.Core features are:
- unlimited soft-breakpoints via code patching
- hardware-assisted x86 breakpoints and watchpoints
Quoted 7 lines
Changes in this version:
- use generic hookcpu_synchronize_stateto transfer registers between user space and kvm
- pushkvm_sw_breakpointsinto KVMStateSigned-off-by Jan Kiszka <[email protected]>
Applied both. Thanks.
Regards,
Anthony Liguori