From: Paolo Bonzini <pbonzini@redhat.com>
To: Avi Kivity <avi@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>,
liu ping fan <qemulist@gmail.com>,
qemu-devel@nongnu.org, Blue Swirl <blauwirbel@gmail.com>,
Alex Williamson <alex.williamson@redhat.com>,
Anthony Liguori <anthony@codemonkey.ws>
Subject: Re: [Qemu-devel] [RFC v1 4/7] pci: switch iommu to using the memory API
Date: Thu, 11 Oct 2012 15:53:13 +0200 [thread overview]
Message-ID: <5076CF49.7080300@redhat.com> (raw)
In-Reply-To: <1349962023-560-5-git-send-email-avi@redhat.com>
Il 11/10/2012 15:27, Avi Kivity ha scritto:
> -static int spapr_tce_translate(DMAContext *dma,
> - dma_addr_t addr,
> - target_phys_addr_t *paddr,
> - target_phys_addr_t *len,
> - DMADirection dir)
> +IOMMUTLBEntry spapr_tce_translate(DMAContext *dma, target_phys_addr_t addr,
> + bool is_write)
> {
> sPAPRTCETable *tcet = DO_UPCAST(sPAPRTCETable, dma, dma);
> - enum sPAPRTCEAccess access = (dir == DMA_DIRECTION_FROM_DEVICE)
> - ? SPAPR_TCE_WO : SPAPR_TCE_RO;
> + enum sPAPRTCEAccess access = is_write ? SPAPR_TCE_WO : SPAPR_TCE_RO;
> uint64_t tce;
>
> #ifdef DEBUG_TCE
> @@ -84,29 +80,27 @@ static int spapr_tce_translate(DMAContext *dma,
> #ifdef DEBUG_TCE
> fprintf(stderr, "spapr_tce_translate out of bounds\n");
> #endif
> - return -EFAULT;
> + return (IOMMUTLBEntry) { .valid = false };
> }
>
> tce = tcet->table[addr >> SPAPR_TCE_PAGE_SHIFT].tce;
>
> /* Check TCE */
> if (!(tce & access)) {
> - return -EPERM;
> + return (IOMMUTLBEntry) { .valid = false };
> }
>
> - /* How much til end of page ? */
> - *len = ((~addr) & SPAPR_TCE_PAGE_MASK) + 1;
> -
> - /* Translate */
> - *paddr = (tce & ~SPAPR_TCE_PAGE_MASK) |
> - (addr & SPAPR_TCE_PAGE_MASK);
> -
> #ifdef DEBUG_TCE
> fprintf(stderr, " -> *paddr=0x" TARGET_FMT_plx ", *len=0x"
> - TARGET_FMT_plx "\n", *paddr, *len);
> + TARGET_FMT_plx "\n", (tce & ~SPAPR_TCE_PAGE_MASK), SPAPR_TCE_PAGE_MASK + 1);
> #endif
>
> - return 0;
> + return (IOMMUTLBEntry) {
> + .device_addr = addr & SPAPR_TCE_PAGE_MASK,
> + .translated_addr = (tce & ~SPAPR_TCE_PAGE_MASK),
> + .addr_mask = SPAPR_TCE_PAGE_MASK,
> + .valid = true,
> + };
> }
>
> DMAContext *spapr_tce_new_dma_context(uint32_t liobn, size_t window_size)
> @@ -118,7 +112,7 @@ DMAContext *spapr_tce_new_dma_context(uint32_t liobn, size_t window_size)
> }
>
> tcet = g_malloc0(sizeof(*tcet));
> - dma_context_init(&tcet->dma, &address_space_memory, spapr_tce_translate, NULL, NULL);
> + dma_context_init(&tcet->dma, &address_space_memory, NULL, NULL, NULL);
>
> tcet->liobn = liobn;
> tcet->window_size = window_size;
> @@ -253,7 +247,8 @@ int spapr_tcet_dma_dt(void *fdt, int node_off, const char *propname,
> return 0;
> }
>
> - if (iommu->translate == spapr_tce_translate) {
> + /* FIXME: WHAT?? */
> + if (true) {
> sPAPRTCETable *tcet = DO_UPCAST(sPAPRTCETable, dma, iommu);
> return spapr_dma_dt(fdt, node_off, propname,
> tcet->liobn, 0, tcet->window_size);
> diff --git a/hw/spapr_pci.c b/hw/spapr_pci.c
> index 661c05b..24f5b46 100644
> --- a/hw/spapr_pci.c
> +++ b/hw/spapr_pci.c
> @@ -506,14 +506,30 @@ static void spapr_msi_write(void *opaque, target_phys_addr_t addr,
> /*
> * PHB PCI device
> */
> -static DMAContext *spapr_pci_dma_context_fn(PCIBus *bus, void *opaque,
> - int devfn)
> +static MemoryRegion *spapr_pci_dma_iommu_new(PCIBus *bus, void *opaque, int devfn)
> {
> sPAPRPHBState *phb = opaque;
>
> - return phb->dma;
> + return &phb->iommu;
> }
>
> +static void spapr_pci_dma_iommu_destroy(MemoryRegion *iommu)
> +{
> + /* iommu is shared among devices, do nothing */
> +}
> +
> +static IOMMUTLBEntry spapr_phb_translate(MemoryRegion *iommu, target_phys_addr_t addr,
> + bool is_write)
> +{
> + sPAPRPHBState *phb = container_of(iommu, sPAPRPHBState, iommu);
> +
> + return spapr_tce_translate(phb->dma, addr, is_write);
> +}
IIUC, sPAPRTCETable could drop the DMAContext field and just include the
actual translation info. spapr_tce_new_dma_context becomes
spapr_tce_new and returns an opaque sPAPRTCETable struct that is passed
back to spapr_tce_translate.
And DMAContext can disappear and will be replaced with just an
AddressSpace *.
I like it!
Paolo
next prev parent reply other threads:[~2012-10-11 13:53 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-10-11 13:26 [Qemu-devel] [RFC v1 0/7] IOMMU support Avi Kivity
2012-10-11 13:26 ` [Qemu-devel] [RFC v1 1/7] memory: fix address space initialization/destruction Avi Kivity
2012-10-11 13:31 ` Paolo Bonzini
2012-10-11 13:33 ` Avi Kivity
2012-10-13 9:14 ` Blue Swirl
2012-10-11 13:26 ` [Qemu-devel] [RFC v1 2/7] memory: limit sections in the radix tree to the actual address space size Avi Kivity
2012-10-11 13:26 ` [Qemu-devel] [RFC v1 3/7] memory: iommu support Avi Kivity
2012-10-11 13:42 ` Paolo Bonzini
2012-10-11 13:45 ` Avi Kivity
2012-10-11 13:54 ` Paolo Bonzini
2012-10-11 13:57 ` Avi Kivity
2012-10-12 2:51 ` Benjamin Herrenschmidt
2012-10-15 16:54 ` Avi Kivity
2012-10-12 2:45 ` Benjamin Herrenschmidt
2012-10-13 9:30 ` Blue Swirl
2012-10-13 11:37 ` Benjamin Herrenschmidt
2012-10-11 14:29 ` Avi Kivity
2012-10-11 13:27 ` [Qemu-devel] [RFC v1 4/7] pci: switch iommu to using the memory API Avi Kivity
2012-10-11 13:53 ` Paolo Bonzini [this message]
2012-10-11 13:56 ` Avi Kivity
2012-10-13 9:13 ` Blue Swirl
2012-10-15 10:31 ` Avi Kivity
2012-10-11 13:27 ` [Qemu-devel] [RFC v1 5/7] i440fx: add an iommu Avi Kivity
2012-10-11 13:27 ` [Qemu-devel] [RFC v1 6/7] vfio: abort if an emulated iommu is used Avi Kivity
2012-10-11 13:27 ` [Qemu-devel] [RFC v1 7/7] vhost: " Avi Kivity
2012-10-11 13:31 ` Michael S. Tsirkin
2012-10-11 13:34 ` Avi Kivity
2012-10-11 13:44 ` Michael S. Tsirkin
2012-10-11 13:44 ` Avi Kivity
2012-10-11 14:35 ` Michael S. Tsirkin
2012-10-11 14:35 ` Avi Kivity
2012-10-11 15:34 ` Michael S. Tsirkin
2012-10-11 15:48 ` Avi Kivity
2012-10-11 19:38 ` Alex Williamson
2012-10-15 10:24 ` Avi Kivity
2012-10-15 8:44 ` liu ping fan
2012-10-15 10:32 ` Avi Kivity
2012-10-12 2:36 ` [Qemu-devel] [RFC v1 0/7] IOMMU support Benjamin Herrenschmidt
2012-10-15 10:45 ` Avi Kivity
2012-10-15 19:52 ` Benjamin Herrenschmidt
2012-10-16 9:30 ` Avi Kivity
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5076CF49.7080300@redhat.com \
--to=pbonzini@redhat.com \
--cc=alex.williamson@redhat.com \
--cc=anthony@codemonkey.ws \
--cc=avi@redhat.com \
--cc=blauwirbel@gmail.com \
--cc=mst@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=qemulist@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).