Re: [Qemu-devel] [V18 2/4] hw/i386/trace-events: Add AMD IOMMU trace events

[David Kiarie]

On Tue, Sep 20, 2016 at 4:40 AM, Michael S. Tsirkin <address@hidden> wrote:

> On Tue, Sep 20, 2016 at 04:14:14AM +0300, David Kiarie wrote:
> >
> >
> > On Tue, Sep 20, 2016 at 3:09 AM, Michael S. Tsirkin <address@hidden>
> wrote:
> >
> >     On Tue, Sep 20, 2016 at 03:05:02AM +0300, David Kiarie wrote:
> >     > Signed-off-by: David Kiarie <address@hidden>
> >     > ---
> >     >  hw/i386/trace-events | 29 +++++++++++++++++++++++++++++
> >     >  1 file changed, 29 insertions(+)
> >     >
> >     > diff --git a/hw/i386/trace-events b/hw/i386/trace-events
> >     > index 5b99eba..ddeda02 100644
> >     > --- a/hw/i386/trace-events
> >     > +++ b/hw/i386/trace-events
> >     > @@ -13,3 +13,32 @@ mhp_pc_dimm_assigned_address(uint64_t addr)
> >     "0x%"PRIx64
> >     >
> >     >  # hw/i386/x86-iommu.c
> >     >  x86_iommu_iec_notify(bool global, uint32_t index, uint32_t mask)
> "Notify
> >     IEC invalidation: global=%d index=%" PRIu32 " mask=%" PRIu32
> >     > +
> >     > +# hw/i386/amd_iommu.c
> >     > +amdvi_evntlog_fail(uint64_t addr, uint32_t head) "error: fail to
> write
> >     at addr 0x%"PRIx64 " +  offset 0x%"PRIx32
> >     > +amdvi_cache_update(uint16_t domid, uint32_t bus, uint32_t slot,
> uint32_t
> >     func, uint64_t gpa, uint64_t txaddr) " update iotlb domid 0x%"PRIx16"
> >     devid: %02x:%02x.%x gpa 0x%"PRIx64 " hpa 0x%"PRIx64
> >     > +amdvi_completion_wait_fail(uint64_t addr) "error: fail to write
> at
> >     address 0x%"PRIx64
> >     > +amdvi_mmio_write(const char *reg, uint64_t addr, unsigned size,
> uint64_t
> >     val, unsigned long offset) "%s write addr 0x%"PRIx64 ", size %d, val
> >     0x%"PRIx64 ", offset 0x%"PRIx64
> >
> >
> >     unsigned long offset should be printed with %0lx not PRIx64.
> >
> >     There are a bunch of other bugs I fixed silenty in this file,
> >     I don't remember where exactly pls go over and and fix
> >     all you can find.
> >
> >
> >
> > Separately resending this patch in the meantime just in case the rest of
> the
> > patchset happens not to have any more issues.
> >
> >
>

Hi  Michael,

>     > +amdvi_mmio_read(const char *reg, uint64_t addr, unsigned size,
> uint64_t
> >     offset) "%s read addr 0x%"PRIx64", size %d offset 0x%"PRIx64
> >     > +amdvi_command_error(uint64_t status) "error: Executing commands
> with
> >     command buffer disabled 0x%"PRIx64
> >     > +amdvi_command_read_fail(uint64_t addr, uint32_t head) "error:
> fail to
> >     access memory at 0x%"PRIx64" + 0x%"PRIu32
> >     > +amdvi_command_exec(uint32_t head, uint32_t tail, uint64_t buf)
> "command
> >     buffer head at 0x%"PRIx32 " command buffer tail at 0x%"PRIx32"
> command
> >     buffer base at 0x%" PRIx64
> >     > +amdvi_unhandled_command(uint8_t type) "unhandled command %d"
> >     > +amdvi_intr_inval(void) "Interrupt table invalidated"
> >     > +amdvi_iotlb_inval(void) "IOTLB pages invalidated"
> >     > +amdvi_prefetch_pages(void) "Pre-fetch of AMD-Vi pages requested"
> >     > +amdvi_pages_inval(uint16_t domid) "AMD-Vi pages for domain
> 0x%"PRIx16 "
> >     invalidated"
> >     > +amdvi_all_inval(void) "Invalidation of all AMD-Vi cache requested
> "
> >     > +amdvi_ppr_exec(void) "Execution of PPR queue requested "
> >     > +amdvi_devtab_inval(uint16_t bus, uint16_t slot, uint16_t func)
> "device
> >     table entry for devid: %02x:%02x.%x invalidated"
> >     > +amdvi_completion_wait(uint64_t addr, uint64_t data) "completion
> wait
> >     requested with store address 0x%"PRIx64" and store data 0x%"PRIx64
> >     > +amdvi_control_status(uint64_t val) "MMIO_STATUS state 0x%"PRIx64
> >     > +amdvi_iotlb_reset(void) "IOTLB exceed size limit - reset "
> >     > +amdvi_completion_wait_exec(uint64_t addr, uint64_t data)
> "completion
> >     wait requested with store address 0x%"PRIx64" and store data
> 0x%"PRIx64
> >     > +amdvi_dte_get_fail(uint64_t addr, uint32_t offset) "error: failed
> to
> >     access Device Entry devtab 0x%"PRIx64" offset 0x%"PRIx32
> >     > +amdvi_invalid_dte(uint64_t addr) "PTE entry at 0x%"PRIx64" is
> invalid "
> >     > +amdvi_get_pte_hwerror(uint64_t addr) "hardware error eccessing
> PTE at
> >     addr 0x%"PRIx64
> >     > +amdvi_mode_invalid(unsigned level, uint64_t addr)"error:
> translation
> >     level 0x%"PRIu8" translating addr 0x%"PRIx64
>
> This is also wrong. Can you pls go over the traces?
>

I think you might be looking the wrong patch here.


> Also, you can check Peter Maydell's rejects of the pulls that
> included these patches.


I did look at the reject. The initial code wasn't actually broken but that
discussion wasn't really relevant since 'extract64/extract32' is more
decent.


> >     > +amdvi_page_fault(uint64_t addr) "error: page fault accessing guest
> >     physical address 0x%"PRIx64
> >     > +amdvi_iotlb_hit(uint16_t bus, uint16_t slot, uint16_t func,
> uint64_t
> >     addr, uint64_t txaddr) "hit iotlb devid %02x:%02x.%x gpa 0x%"PRIx64
> " hpa
> >     0x%"PRIx64
> >     > +amdvi_translation_result(uint16_t bus, uint16_t slot, uint16_t
> func,
> >     uint64_t addr, uint64_t txaddr) "devid: %02x:%02x.%x gpa 0x%"PRIx64
> " hpa
> >     0x%"PRIx64
> >     > --
> >     > 2.1.4
> >
> >
>