[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH v2] intel_iommu: relax iq tail check on VTD_GCMD
From: |
Peter Xu |
Subject: |
Re: [Qemu-devel] [PATCH v2] intel_iommu: relax iq tail check on VTD_GCMD_QIE enable |
Date: |
Mon, 19 Jun 2017 16:29:05 +0800 |
User-agent: |
Mutt/1.5.24 (2015-08-30) |
On Mon, Jun 19, 2017 at 09:31:16AM +0200, Ladi Prosek wrote:
> The VT-d spec (section 6.5.2) prescribes software to zero the
> Invalidation Queue Tail Register before enabling the VTD_GCMD_QIE
> Global Command Register bit. Windows Server 2012 R2 and possibly
> other older Windows versions violate the protocol and set a
> non-zero queue tail first, which in effect makes them crash early
> on boot with -device intel-iommu,intremap=on.
>
> This commit relaxes the check and instead of failing to enable
> VTD_GCMD_QIE with vtd_err_qi_enable, it behaves as if the tail
> register was set just after enabling VTD_GCMD_QIE
> (see vtd_handle_iqt_write).
>
> Signed-off-by: Ladi Prosek <address@hidden>
Reviewed-by: Peter Xu <address@hidden>
> ---
>
> v1->v2:
> * rebased on top of Michael's repo
> * trace_vtd_warn_invalid_qi_tail instead of VTD_DPRINTF, called on
> non-zero iq_tail
> * added a comment
>
>
> hw/i386/intel_iommu.c | 33 +++++++++++++++++++--------------
> hw/i386/trace-events | 2 +-
> 2 files changed, 20 insertions(+), 15 deletions(-)
>
> diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
> index a9b59bd..2ddf3bd 100644
> --- a/hw/i386/intel_iommu.c
> +++ b/hw/i386/intel_iommu.c
> @@ -1450,10 +1450,7 @@ static uint64_t vtd_iotlb_flush(IntelIOMMUState *s,
> uint64_t val)
> return iaig;
> }
>
> -static inline bool vtd_queued_inv_enable_check(IntelIOMMUState *s)
> -{
> - return s->iq_tail == 0;
> -}
> +static void vtd_fetch_inv_desc(IntelIOMMUState *s);
>
> static inline bool vtd_queued_inv_disable_check(IntelIOMMUState *s)
> {
> @@ -1468,16 +1465,24 @@ static void vtd_handle_gcmd_qie(IntelIOMMUState *s,
> bool en)
> trace_vtd_inv_qi_enable(en);
>
> if (en) {
> - if (vtd_queued_inv_enable_check(s)) {
> - s->iq = iqa_val & VTD_IQA_IQA_MASK;
> - /* 2^(x+8) entries */
> - s->iq_size = 1UL << ((iqa_val & VTD_IQA_QS) + 8);
> - s->qi_enabled = true;
> - trace_vtd_inv_qi_setup(s->iq, s->iq_size);
> - /* Ok - report back to driver */
> - vtd_set_clear_mask_long(s, DMAR_GSTS_REG, 0, VTD_GSTS_QIES);
> - } else {
> - trace_vtd_err_qi_enable(s->iq_tail);
> + s->iq = iqa_val & VTD_IQA_IQA_MASK;
> + /* 2^(x+8) entries */
> + s->iq_size = 1UL << ((iqa_val & VTD_IQA_QS) + 8);
> + s->qi_enabled = true;
> + trace_vtd_inv_qi_setup(s->iq, s->iq_size);
> + /* Ok - report back to driver */
> + vtd_set_clear_mask_long(s, DMAR_GSTS_REG, 0, VTD_GSTS_QIES);
> +
> + if (s->iq_tail != 0) {
> + /*
> + * This is a spec violation but Windows guests are known to set
> up
> + * Queued Invalidation this way so we allow the write and process
> + * Invalidation Descriptors right away.
> + */
> + trace_vtd_warn_invalid_qi_tail(s->iq_tail);
> + if (!(vtd_get_long_raw(s, DMAR_FSTS_REG) & VTD_FSTS_IQE)) {
> + vtd_fetch_inv_desc(s);
> + }
> }
> } else {
> if (vtd_queued_inv_disable_check(s)) {
> diff --git a/hw/i386/trace-events b/hw/i386/trace-events
> index 5f111d6..42d8a7e 100644
> --- a/hw/i386/trace-events
> +++ b/hw/i386/trace-events
> @@ -74,7 +74,7 @@ vtd_err_dmar_slpte_read_error(uint64_t iova, int level)
> "iova 0x%"PRIx64" level
> vtd_err_dmar_slpte_perm_error(uint64_t iova, int level, uint64_t slpte, bool
> is_write) "iova 0x%"PRIx64" level %d slpte 0x%"PRIx64" write %d"
> vtd_err_dmar_slpte_resv_error(uint64_t iova, int level, uint64_t slpte)
> "iova 0x%"PRIx64" level %d slpte 0x%"PRIx64
> vtd_err_dmar_translate(uint8_t bus, uint8_t slot, uint8_t func, uint64_t
> iova) "dev %02x:%02x.%02x iova 0x%"PRIx64
> -vtd_err_qi_enable(uint16_t tail) "tail 0x%"PRIx16
> +vtd_warn_invalid_qi_tail(uint16_t tail) "tail 0x%"PRIx16
> vtd_err_qi_disable(uint16_t head, uint16_t tail, int type) "head 0x%"PRIx16"
> tail 0x%"PRIx16" last_desc_type %d"
> vtd_err_qi_tail(uint16_t tail, uint16_t size) "tail 0x%"PRIx16" size
> 0x%"PRIx16
> vtd_err_irte(int index, uint64_t lo, uint64_t hi) "index %d low 0x%"PRIx64"
> high 0x%"PRIx64
> --
> 2.9.3
>
--
Peter Xu