ide/atapi: Mark non-data commands as complete
PATCH 1/1 ide/atapi: Mark non-data commands as complete
When the command completion code in IDE and AHCI was unified to put all command completion inside of a callback, "cmd_done," we neglected to ensure that all AHCI/ATAPI command paths would eventually register as finished. for the PCI interface to IDE this is not a problem because cmd_done is a nop, but the AHCI implementation needs to send a D2H_REG_FIS and interrupt back to the guest to inform of completion.
This patch adds calls to ide_set_inactive, which calls ide_cmd_done, inside of ide_atapi_cmd_ok and ide_atapi_cmd_error.
This fixes regressions observed by trying to boot QEMU with a Fedora 20 live CD under Q35/AHCI, which uses ATAPI command 0x00, which is a status check that may cause a hang because we never complete, and ATAPI command 0x56, which is unsupported by our current implementation and results in an error that we never report back to the guest.
--- hw/ide/atapi.c | 2 ++ 1 file changed, 2 insertions(+)
hw/ide/atapi.c +2 -0
@@ -134,6 +134,7 @@ void ide_atapi_cmd_ok(IDEState *s) s->error =0 ; s->status =READY_STAT |SEEK_STAT ; s->nsector = (s->nsector & ~7 ) |ATAPI_INT_REASON_IO |ATAPI_INT_REASON_CD ; +ide_set_inactive (s, false);ide_set_irq (s->bus); }
@@ -147,6 +148,7 @@ void ide_atapi_cmd_error(IDEState *s, int sense_key, int asc) s->nsector = (s->nsector & ~7 ) |ATAPI_INT_REASON_IO |ATAPI_INT_REASON_CD ; s->sense_key = sense_key; s->asc = asc; +ide_set_inactive (s, false);ide_set_irq (s->bus); }
Signature · 1 line
1.9.3
Il 12/09/2014 18:00, John Snow ha scritto:
Quoted 41 lines
When the command completion code in IDE and AHCI was unified to put all command completion inside of a callback, "
cmd_done," we neglected to ensure that all AHCI/ATAPI command paths would eventually register as finished. for the PCI interface to IDE this is not a problem becausecmd_doneis a nop, but the AHCI implementation needs to send aD2H_REG_FISand interrupt back to the guest to inform of completion.This patch adds calls to
ide_set_inactive, which callside_cmd_done, inside ofide_atapi_cmd_okandide_atapi_cmd_error.This fixes regressions observed by trying to boot QEMU with a Fedora 20 live CD under Q35/AHCI, which uses ATAPI command
0x00, which is a status check that may cause a hang because we never complete, and ATAPI command0x56, which is unsupported by our current implementation and results in an error that we never report back to the guest.Signed-off-by John Snow <[email protected]>--- hw/ide/atapi.c | 2 ++ 1 file changed, 2 insertions(+)hw/ide/atapi.c +1
-0@@ -134,6 +134,7 @@ void ide_atapi_cmd_ok(IDEState *s) s->error =0 ; s->status =READY_STAT |SEEK_STAT ; s->nsector = (s->nsector & ~7 ) |ATAPI_INT_REASON_IO |ATAPI_INT_REASON_CD ; +ide_set_inactive (s, false);ide_set_irq (s->bus); }
hw/ide/atapi.c void ide_atapi_cmd_error(IDEState *s, int sense_key, int asc) · line 154
Quoted 7 lines
hw/ide/atapi.c +1
-0@@ -147,6 +148,7 @@ void ide_atapi_cmd_error(IDEState *s, int sense_key, int asc) s->nsector = (s->nsector & ~7 ) |ATAPI_INT_REASON_IO |ATAPI_INT_REASON_CD ; s->sense_key = sense_key; s->asc = asc; +ide_set_inactive (s, false);ide_set_irq (s->bus); }
The set_inactive callback does nothing on AHCI, but is DMA-specific for PCI. Even though you probably aren't seeing any bad effects, I think ide_transfer_stop is a better match. It would also match what ide_atapi_cmd_reply_end does before calling ide_atapi_cmd_error (via ide_atapi_io_error) and before an inlined copy of ide_atapi_cmd_ok. So you would get a bugfix and a cleanup at once.
Paolo
Quoted 57 lines · Paolo Bonzini
Il 12/09/2014 18:00, John Snow ha scritto:
Quoted 41 lines
When the command completion code in IDE and AHCI was unified to put all command completion inside of a callback, "
cmd_done," we neglected to ensure that all AHCI/ATAPI command paths would eventually register as finished. for the PCI interface to IDE this is not a problem becausecmd_doneis a nop, but the AHCI implementation needs to send aD2H_REG_FISand interrupt back to the guest to inform of completion.This patch adds calls to
ide_set_inactive, which callside_cmd_done, inside ofide_atapi_cmd_okandide_atapi_cmd_error.This fixes regressions observed by trying to boot QEMU with a Fedora 20 live CD under Q35/AHCI, which uses ATAPI command
0x00, which is a status check that may cause a hang because we never complete, and ATAPI command0x56, which is unsupported by our current implementation and results in an error that we never report back to the guest.Signed-off-by John Snow <[email protected]>--- hw/ide/atapi.c | 2 ++ 1 file changed, 2 insertions(+)hw/ide/atapi.c +1
-0@@ -134,6 +134,7 @@ void ide_atapi_cmd_ok(IDEState *s) s->error =0 ; s->status =READY_STAT |SEEK_STAT ; s->nsector = (s->nsector & ~7 ) |ATAPI_INT_REASON_IO |ATAPI_INT_REASON_CD ; +ide_set_inactive (s, false);ide_set_irq (s->bus); }hw/ide/atapi.c void ide_atapi_cmd_error(IDEState *s, int sense_key, int asc) · line 154
Quoted 7 lines
hw/ide/atapi.c +1
-0@@ -147,6 +148,7 @@ void ide_atapi_cmd_error(IDEState *s, int sense_key, int asc) s->nsector = (s->nsector & ~7 ) |ATAPI_INT_REASON_IO |ATAPI_INT_REASON_CD ; s->sense_key = sense_key; s->asc = asc; +ide_set_inactive (s, false);ide_set_irq (s->bus); }The
set_inactivecallback does nothing on AHCI, but is DMA-specific for PCI. Even though you probably aren't seeing any bad effects, I thinkide_transfer_stopis a better match. It would also match whatide_atapi_cmd_reply_enddoes before callingide_atapi_cmd_error(viaide_atapi_io_error) and before an inlined copy ofide_atapi_cmd_ok. So you would get a bugfix and a cleanup at once.
Quoted 2 lines
Paolo
OK. I am going to hold on to the revised patch for extended testing, I am seeing some strange issues and I want to fix it correctly instead of introduce new regressions with a quick fix.
--j