Re: [Qemu-devel] [PATCH 24/26] acpi, acpi_piix: factor out GPE logic
On Sun, Apr 17, 2011 at 06:53:12PM +0300, Avi Kivity wrote: On 04/17/2011 04:50 PM, Isaku Yamahata wrote: On Sun, Apr 17, 2011 at 04:17:51PM +0300, Avi Kivity wrote: On 03/16/2011 11:29 AM, Isaku Yamahata wrote: factor out ACPI GPE logic. Later it will be used by ICH9 ACPI. I think this patch is causing qemu-kvm failures on migration: (gdb) bt #0 0x0049aff4 in qemu_put_be16s (f=0x1a74490, pv=0x2c02580, size=2) at hw/hw.h:108 #1 put_uint16 (f=0x1a74490, pv=0x2c02580, size=2) at savevm.c:855 #2 0x0049c3e4 in vmstate_save_state (f=0x1a74490, vmsd=0x6f0b00, opaque=0x1842ef0) at savevm.c:1436 #3 0x0049c3b6 in vmstate_save_state (f=0x1a74490, vmsd=0x6f0aa0, opaque=0x1842b90) at savevm.c:1434 #4 0x0049c6f1 in vmstate_save (mon=value optimized out, f=0x1a74490) at savevm.c:1459 #5 qemu_savevm_state_complete (mon=value optimized out, f=0x1a74490) at savevm.c:1600 #6 0x0049455a in migrate_fd_put_ready (opaque=0x1847890) at migration.c:383 #7 0x004ce2eb in qemu_run_timers (clock=value optimized out) at qemu-timer.c:505 #8 0x004ce806 in qemu_run_all_timers () at qemu-timer.c:619 #9 0x00419463 in main_loop_wait (nonblocking=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1339 #10 0x00433927 in kvm_main_loop () at /build/home/tlv/akivity/qemu-kvm/qemu-kvm.c:1590 #11 0x0041a3a6 in main_loop (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1369 #12 main (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:3257 The vmstate being migrated is gpe. +#define VMSTATE_GPE_ARRAY(_field, _state)\ + { \ + .name = (stringify(_field)), \ + .version_id = 0,\ + .num= GPE_LEN, \ + .info =vmstate_info_uint16, \ + .size = sizeof(uint16_t), \ + .flags = VMS_ARRAY | VMS_POINTER, \ + .offset = vmstate_offset_pointer(_state, _field, uint8_t), \ + } + static const VMStateDescription vmstate_gpe = { .name = gpe, .version_id = 1, .minimum_version_id = 1, .minimum_version_id_old = 1, .fields = (VMStateField []) { -VMSTATE_UINT16(sts, struct gpe_regs), -VMSTATE_UINT16(en, struct gpe_regs), +VMSTATE_GPE_ARRAY(sts, ACPIGPE), +VMSTATE_GPE_ARRAY(en, ACPIGPE), VMSTATE_END_OF_LIST() } }; I'm no vmstate expert, but this does look odd. Why both VMS_ARRAY and VMS_POINTER? aren't we trying to save/restore a simple 16-bit value? Or at least we did before this patch. That's right. the difference is, the new member type became uint8_t*. Does the following help? Unforunately not. The core is very similar though: Hmm but it seems that there is another issue. With the previous patch, I confirmed sts and en are saved as expected. The related VMStateDescription are defined in acpi_piix.c in the following order. vmstate_acpi calls vmstate_gpe. and vmstate_gpe vmstate_acpi. However, your call trace shows something different. (I'm assuming the stack trace is correct. gdb or compiler optimization?) static const VMStateDescription vmstate_gpe = { ... static const VMStateDescription vmstate_acpi = { .fields = (VMStateField []) { ... VMSTATE_STRUCT(gpe, PIIX4PMState, 2, vmstate_gpe, ACPIGPE), #0 0x0049aff4 in qemu_put_be16s (f=0x2427490, pv=0x1ae7580, size=2) at hw/hw.h:108 #1 put_uint16 (f=0x2427490, pv=0x1ae7580, size=2) at savevm.c:855 #2 0x0049c3e4 in vmstate_save_state (f=0x2427490, vmsd=0x6f0b00, opaque=0x21f5ef0) at savevm.c:1436 vmsd = 0x6f0b00 = vmstate_gpe correct??? opaque = 0x21f5ef0 = PIIX4PMState-gpe #3 0x0049c3b6 in vmstate_save_state (f=0x2427490, vmsd=0x6f0aa0, opaque=0x21f5b90) at savevm.c:1434 vmsd = 0x6f0aa0 = vmstate_acpi correct? opaque = 0x21f5b90 = PIIX4PMState opaque seems correct as 0x21f5b90 - 0x21f5ef0 = 0x348 = offsetof(PIIX4PMState, gpe) On my environment the stack trace looks like (gdb) where #0 vmstate_save_state (f=0x25d16d0, vmsd=0x695940, opaque=0x283a218) at savevm.c:1374 #1 0x004bb4f7 in vmstate_save_state (f=0x25d16d0, vmsd=0x695a40, opaque=0x2839ed0) at savevm.c:1412 ... (gdb) print vmstate_gpe $32 = (const VMStateDescription *) 0x695940 (gdb) print vmstate_acpi $33 = (const VMStateDescription *) 0x695a40 thanks, #4 0x0049c6f1 in vmstate_save
Re: [Qemu-devel] [PATCH 24/26] acpi, acpi_piix: factor out GPE logic
On 04/18/2011 10:47 AM, Isaku Yamahata wrote: Hmm but it seems that there is another issue. With the previous patch, I confirmed sts and en are saved as expected. The related VMStateDescription are defined in acpi_piix.c in the following order. vmstate_acpi calls vmstate_gpe. andvmstate_gpe vmstate_acpi. (gdb) p vmstate_gpe $1 = (const VMStateDescription *) 0x6f0b00 (gdb) p vmstate_acpi $2 = (const VMStateDescription *) 0x6f0aa0 So no. However, your call trace shows something different. (I'm assuming the stack trace is correct. gdb or compiler optimization?) static const VMStateDescription vmstate_gpe = { ... static const VMStateDescription vmstate_acpi = { .fields = (VMStateField []) { ... VMSTATE_STRUCT(gpe, PIIX4PMState, 2, vmstate_gpe, ACPIGPE), #0 0x0049aff4 in qemu_put_be16s (f=0x2427490, pv=0x1ae7580, size=2) at hw/hw.h:108 #1 put_uint16 (f=0x2427490, pv=0x1ae7580, size=2) at savevm.c:855 #2 0x0049c3e4 in vmstate_save_state (f=0x2427490, vmsd=0x6f0b00, opaque=0x21f5ef0) at savevm.c:1436 vmsd = 0x6f0b00 =vmstate_gpe correct??? Correct. opaque = 0x21f5ef0 =PIIX4PMState-gpe #3 0x0049c3b6 in vmstate_save_state (f=0x2427490, vmsd=0x6f0aa0, opaque=0x21f5b90) at savevm.c:1434 vmsd = 0x6f0aa0 =vmstate_acpi correct? Correct. opaque = 0x21f5b90 =PIIX4PMState opaque seems correct as 0x21f5b90 - 0x21f5ef0 = 0x348 = offsetof(PIIX4PMState, gpe) On my environment the stack trace looks like (gdb) where #0 vmstate_save_state (f=0x25d16d0, vmsd=0x695940, opaque=0x283a218) at savevm.c:1374 #1 0x004bb4f7 in vmstate_save_state (f=0x25d16d0, vmsd=0x695a40, opaque=0x2839ed0) at savevm.c:1412 ... (gdb) printvmstate_gpe $32 = (const VMStateDescription *) 0x695940 (gdb) printvmstate_acpi $33 = (const VMStateDescription *) 0x695a40 Are you using qemu-kvm.git or qemu.git? I think there are indeed two issues, your patch fixes the first and there is another that is specific to qemu-kvm. -- I have a truly marvellous patch that fixes the bug which this signature is too narrow to contain.
Re: [Qemu-devel] [PATCH 24/26] acpi, acpi_piix: factor out GPE logic
On Mon, Apr 18, 2011 at 11:22:40AM +0300, Avi Kivity wrote: Are you using qemu-kvm.git or qemu.git? I think there are indeed two issues, your patch fixes the first and there is another that is specific to qemu-kvm. I'm using qemu.git. I've start to have a look at qemu-kvm.git, -- yamahata
Re: [Qemu-devel] [PATCH 24/26] acpi, acpi_piix: factor out GPE logic
On 03/16/2011 11:29 AM, Isaku Yamahata wrote: factor out ACPI GPE logic. Later it will be used by ICH9 ACPI. I think this patch is causing qemu-kvm failures on migration: (gdb) bt #0 0x0049aff4 in qemu_put_be16s (f=0x1a74490, pv=0x2c02580, size=2) at hw/hw.h:108 #1 put_uint16 (f=0x1a74490, pv=0x2c02580, size=2) at savevm.c:855 #2 0x0049c3e4 in vmstate_save_state (f=0x1a74490, vmsd=0x6f0b00, opaque=0x1842ef0) at savevm.c:1436 #3 0x0049c3b6 in vmstate_save_state (f=0x1a74490, vmsd=0x6f0aa0, opaque=0x1842b90) at savevm.c:1434 #4 0x0049c6f1 in vmstate_save (mon=value optimized out, f=0x1a74490) at savevm.c:1459 #5 qemu_savevm_state_complete (mon=value optimized out, f=0x1a74490) at savevm.c:1600 #6 0x0049455a in migrate_fd_put_ready (opaque=0x1847890) at migration.c:383 #7 0x004ce2eb in qemu_run_timers (clock=value optimized out) at qemu-timer.c:505 #8 0x004ce806 in qemu_run_all_timers () at qemu-timer.c:619 #9 0x00419463 in main_loop_wait (nonblocking=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1339 #10 0x00433927 in kvm_main_loop () at /build/home/tlv/akivity/qemu-kvm/qemu-kvm.c:1590 #11 0x0041a3a6 in main_loop (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1369 #12 main (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:3257 The vmstate being migrated is gpe. +#define VMSTATE_GPE_ARRAY(_field, _state)\ + { \ + .name = (stringify(_field)), \ + .version_id = 0,\ + .num= GPE_LEN, \ + .info =vmstate_info_uint16, \ + .size = sizeof(uint16_t), \ + .flags = VMS_ARRAY | VMS_POINTER, \ + .offset = vmstate_offset_pointer(_state, _field, uint8_t), \ + } + static const VMStateDescription vmstate_gpe = { .name = gpe, .version_id = 1, .minimum_version_id = 1, .minimum_version_id_old = 1, .fields = (VMStateField []) { -VMSTATE_UINT16(sts, struct gpe_regs), -VMSTATE_UINT16(en, struct gpe_regs), +VMSTATE_GPE_ARRAY(sts, ACPIGPE), +VMSTATE_GPE_ARRAY(en, ACPIGPE), VMSTATE_END_OF_LIST() } }; I'm no vmstate expert, but this does look odd. Why both VMS_ARRAY and VMS_POINTER? aren't we trying to save/restore a simple 16-bit value? Or at least we did before this patch. -- error compiling committee.c: too many arguments to function
Re: [Qemu-devel] [PATCH 24/26] acpi, acpi_piix: factor out GPE logic
On Sun, Apr 17, 2011 at 04:17:51PM +0300, Avi Kivity wrote: On 03/16/2011 11:29 AM, Isaku Yamahata wrote: factor out ACPI GPE logic. Later it will be used by ICH9 ACPI. I think this patch is causing qemu-kvm failures on migration: (gdb) bt #0 0x0049aff4 in qemu_put_be16s (f=0x1a74490, pv=0x2c02580, size=2) at hw/hw.h:108 #1 put_uint16 (f=0x1a74490, pv=0x2c02580, size=2) at savevm.c:855 #2 0x0049c3e4 in vmstate_save_state (f=0x1a74490, vmsd=0x6f0b00, opaque=0x1842ef0) at savevm.c:1436 #3 0x0049c3b6 in vmstate_save_state (f=0x1a74490, vmsd=0x6f0aa0, opaque=0x1842b90) at savevm.c:1434 #4 0x0049c6f1 in vmstate_save (mon=value optimized out, f=0x1a74490) at savevm.c:1459 #5 qemu_savevm_state_complete (mon=value optimized out, f=0x1a74490) at savevm.c:1600 #6 0x0049455a in migrate_fd_put_ready (opaque=0x1847890) at migration.c:383 #7 0x004ce2eb in qemu_run_timers (clock=value optimized out) at qemu-timer.c:505 #8 0x004ce806 in qemu_run_all_timers () at qemu-timer.c:619 #9 0x00419463 in main_loop_wait (nonblocking=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1339 #10 0x00433927 in kvm_main_loop () at /build/home/tlv/akivity/qemu-kvm/qemu-kvm.c:1590 #11 0x0041a3a6 in main_loop (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1369 #12 main (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:3257 The vmstate being migrated is gpe. +#define VMSTATE_GPE_ARRAY(_field, _state)\ + { \ + .name = (stringify(_field)), \ + .version_id = 0,\ + .num= GPE_LEN, \ + .info =vmstate_info_uint16, \ + .size = sizeof(uint16_t), \ + .flags = VMS_ARRAY | VMS_POINTER, \ + .offset = vmstate_offset_pointer(_state, _field, uint8_t), \ + } + static const VMStateDescription vmstate_gpe = { .name = gpe, .version_id = 1, .minimum_version_id = 1, .minimum_version_id_old = 1, .fields = (VMStateField []) { -VMSTATE_UINT16(sts, struct gpe_regs), -VMSTATE_UINT16(en, struct gpe_regs), +VMSTATE_GPE_ARRAY(sts, ACPIGPE), +VMSTATE_GPE_ARRAY(en, ACPIGPE), VMSTATE_END_OF_LIST() } }; I'm no vmstate expert, but this does look odd. Why both VMS_ARRAY and VMS_POINTER? aren't we trying to save/restore a simple 16-bit value? Or at least we did before this patch. That's right. the difference is, the new member type became uint8_t*. Does the following help? diff --git a/hw/acpi_piix4.c b/hw/acpi_piix4.c index 96f5222..3a8fece 100644 --- a/hw/acpi_piix4.c +++ b/hw/acpi_piix4.c @@ -214,10 +214,9 @@ static int vmstate_acpi_post_load(void *opaque, int version_id) { \ .name = (stringify(_field)), \ .version_id = 0,\ - .num= GPE_LEN, \ .info = vmstate_info_uint16, \ .size = sizeof(uint16_t), \ - .flags = VMS_ARRAY | VMS_POINTER, \ + .flags = VMS_SINGLE | VMS_POINTER, \ .offset = vmstate_offset_pointer(_state, _field, uint8_t), \ } -- yamahata
Re: [Qemu-devel] [PATCH 24/26] acpi, acpi_piix: factor out GPE logic
On 04/17/2011 04:50 PM, Isaku Yamahata wrote: On Sun, Apr 17, 2011 at 04:17:51PM +0300, Avi Kivity wrote: On 03/16/2011 11:29 AM, Isaku Yamahata wrote: factor out ACPI GPE logic. Later it will be used by ICH9 ACPI. I think this patch is causing qemu-kvm failures on migration: (gdb) bt #0 0x0049aff4 in qemu_put_be16s (f=0x1a74490, pv=0x2c02580, size=2) at hw/hw.h:108 #1 put_uint16 (f=0x1a74490, pv=0x2c02580, size=2) at savevm.c:855 #2 0x0049c3e4 in vmstate_save_state (f=0x1a74490, vmsd=0x6f0b00, opaque=0x1842ef0) at savevm.c:1436 #3 0x0049c3b6 in vmstate_save_state (f=0x1a74490, vmsd=0x6f0aa0, opaque=0x1842b90) at savevm.c:1434 #4 0x0049c6f1 in vmstate_save (mon=value optimized out, f=0x1a74490) at savevm.c:1459 #5 qemu_savevm_state_complete (mon=value optimized out, f=0x1a74490) at savevm.c:1600 #6 0x0049455a in migrate_fd_put_ready (opaque=0x1847890) at migration.c:383 #7 0x004ce2eb in qemu_run_timers (clock=value optimized out) at qemu-timer.c:505 #8 0x004ce806 in qemu_run_all_timers () at qemu-timer.c:619 #9 0x00419463 in main_loop_wait (nonblocking=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1339 #10 0x00433927 in kvm_main_loop () at /build/home/tlv/akivity/qemu-kvm/qemu-kvm.c:1590 #11 0x0041a3a6 in main_loop (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1369 #12 main (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:3257 The vmstate being migrated is gpe. +#define VMSTATE_GPE_ARRAY(_field, _state)\ + { \ + .name = (stringify(_field)), \ + .version_id = 0,\ + .num= GPE_LEN, \ + .info =vmstate_info_uint16, \ + .size = sizeof(uint16_t), \ + .flags = VMS_ARRAY | VMS_POINTER, \ + .offset = vmstate_offset_pointer(_state, _field, uint8_t), \ + } + static const VMStateDescription vmstate_gpe = { .name = gpe, .version_id = 1, .minimum_version_id = 1, .minimum_version_id_old = 1, .fields = (VMStateField []) { -VMSTATE_UINT16(sts, struct gpe_regs), -VMSTATE_UINT16(en, struct gpe_regs), +VMSTATE_GPE_ARRAY(sts, ACPIGPE), +VMSTATE_GPE_ARRAY(en, ACPIGPE), VMSTATE_END_OF_LIST() } }; I'm no vmstate expert, but this does look odd. Why both VMS_ARRAY and VMS_POINTER? aren't we trying to save/restore a simple 16-bit value? Or at least we did before this patch. That's right. the difference is, the new member type became uint8_t*. Does the following help? Unforunately not. The core is very similar though: #0 0x0049aff4 in qemu_put_be16s (f=0x2427490, pv=0x1ae7580, size=2) at hw/hw.h:108 #1 put_uint16 (f=0x2427490, pv=0x1ae7580, size=2) at savevm.c:855 #2 0x0049c3e4 in vmstate_save_state (f=0x2427490, vmsd=0x6f0b00, opaque=0x21f5ef0) at savevm.c:1436 #3 0x0049c3b6 in vmstate_save_state (f=0x2427490, vmsd=0x6f0aa0, opaque=0x21f5b90) at savevm.c:1434 #4 0x0049c6f1 in vmstate_save (mon=value optimized out, f=0x2427490) at savevm.c:1459 #5 qemu_savevm_state_complete (mon=value optimized out, f=0x2427490) at savevm.c:1600 #6 0x0049455a in migrate_fd_put_ready (opaque=0x21fa8c0) at migration.c:383 #7 0x004ce2eb in qemu_run_timers (clock=value optimized out) at qemu-timer.c:505 #8 0x004ce806 in qemu_run_all_timers () at qemu-timer.c:619 #9 0x00419463 in main_loop_wait (nonblocking=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1339 #10 0x00433927 in kvm_main_loop () at /build/home/tlv/akivity/qemu-kvm/qemu-kvm.c:1590 #11 0x0041a3a6 in main_loop (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:1369 #12 main (argc=value optimized out, argv=value optimized out, envp=value optimized out) at /build/home/tlv/akivity/qemu-kvm/vl.c:3257 -- error compiling committee.c: too many arguments to function