From: David Gibson <david@gibson.dropbear.id.au>
To: Alexey Kardashevskiy <aik@ozlabs.ru>
Cc: qemu-devel@nongnu.org, qemu-ppc@nongnu.org
Subject: Re: [Qemu-devel] [PATCH qemu] target-ppc: kvm: make use of KVM_CREATE_SPAPR_TCE_64
Date: Mon, 9 Jan 2017 13:53:49 +1100 [thread overview]
Message-ID: <20170109025349.GF12515@umbus.fritz.box> (raw)
In-Reply-To: <77140e33-0917-80a1-b2f4-5eaefda45a29@ozlabs.ru>
[-- Attachment #1: Type: text/plain, Size: 8624 bytes --]
On Mon, Jan 09, 2017 at 01:38:26PM +1100, Alexey Kardashevskiy wrote:
> On 03/01/17 13:26, David Gibson wrote:
> > On Thu, Dec 22, 2016 at 12:13:12PM +1100, Alexey Kardashevskiy wrote:
> >> KVM_CAP_SPAPR_TCE capability allows creating TCE tables in KVM which
> >> allows having in-kernel acceleration for H_PUT_TCE_xxx hypercalls.
> >> However it only supports 32bit DMA windows at zero bus offset.
> >>
> >> There is a new KVM_CAP_SPAPR_TCE_64 capability which supports 64bit
> >> window size, variable page size and bus offset.
> >>
> >> This makes use of the new capability. The kernel headers are already
> >> updated as the kernel support went in to v4.6.
> >>
> >> Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
> >> ---
> >> target-ppc/kvm_ppc.h | 12 +++++++-----
> >> hw/ppc/spapr_iommu.c | 8 +++++---
> >> target-ppc/kvm.c | 48 +++++++++++++++++++++++++++++++++++++-----------
> >> 3 files changed, 49 insertions(+), 19 deletions(-)
> >>
> >> diff --git a/target-ppc/kvm_ppc.h b/target-ppc/kvm_ppc.h
> >> index bd1d78bfbe..14320c2378 100644
> >> --- a/target-ppc/kvm_ppc.h
> >> +++ b/target-ppc/kvm_ppc.h
> >> @@ -36,8 +36,9 @@ int kvmppc_booke_watchdog_enable(PowerPCCPU *cpu);
> >> #ifndef CONFIG_USER_ONLY
> >> off_t kvmppc_alloc_rma(void **rma);
> >> bool kvmppc_spapr_use_multitce(void);
> >> -void *kvmppc_create_spapr_tce(uint32_t liobn, uint32_t window_size, int *pfd,
> >> - bool need_vfio);
> >> +void *kvmppc_create_spapr_tce(uint32_t liobn, uint32_t page_shift,
> >> + uint64_t bus_offset, uint32_t nb_table,
> >> + int *pfd, bool need_vfio);
> >> int kvmppc_remove_spapr_tce(void *table, int pfd, uint32_t window_size);
> >> int kvmppc_reset_htab(int shift_hint);
> >> uint64_t kvmppc_rma_size(uint64_t current_size, unsigned int hash_shift);
> >> @@ -168,9 +169,10 @@ static inline bool kvmppc_spapr_use_multitce(void)
> >> return false;
> >> }
> >>
> >> -static inline void *kvmppc_create_spapr_tce(uint32_t liobn,
> >> - uint32_t window_size, int *fd,
> >> - bool need_vfio)
> >> +static inline void *kvmppc_create_spapr_tce(uint32_t liobn, uint32_t page_shift,
> >> + uint64_t bus_offset,
> >> + uint32_t nb_table,
> >> + int *pfd, bool need_vfio)
> >> {
> >> return NULL;
> >> }
> >> diff --git a/hw/ppc/spapr_iommu.c b/hw/ppc/spapr_iommu.c
> >> index ae30bbe30f..29c80bb3c8 100644
> >> --- a/hw/ppc/spapr_iommu.c
> >> +++ b/hw/ppc/spapr_iommu.c
> >> @@ -79,15 +79,16 @@ static IOMMUAccessFlags spapr_tce_iommu_access_flags(uint64_t tce)
> >>
> >> static uint64_t *spapr_tce_alloc_table(uint32_t liobn,
> >> uint32_t page_shift,
> >> + uint64_t bus_offset,
> >> uint32_t nb_table,
> >> int *fd,
> >> bool need_vfio)
> >> {
> >> uint64_t *table = NULL;
> >> - uint64_t window_size = (uint64_t)nb_table << page_shift;
> >>
> >> - if (kvm_enabled() && !(window_size >> 32)) {
> >> - table = kvmppc_create_spapr_tce(liobn, window_size, fd, need_vfio);
> >> + if (kvm_enabled()) {
> >
> > This is broken. Previously, if we had a >4GiB window, we'd fall back
> > to managing it in userspace, which would work, albeit slowly. Now, if
> > you have an older kernel which doesn't support KVM_CAP_SPAPR_TCE_64 it
> > will attempt to allocate it in the kernel, and fail completely.
>
>
> No, kvmppc_create_spapr_tce() would return NULL and right after that there
> is a "if (!table)" (it can be seen at the end of this chunk) to handle the
> failure.
Oh, yes, sorry. For some reason I thought there was a return in that
if block.
> >
> >> + table = kvmppc_create_spapr_tce(liobn, page_shift, bus_offset, nb_table,
> >> + fd, need_vfio);
> >> }
> >>
> >> if (!table) {
> >> @@ -342,6 +343,7 @@ void spapr_tce_table_enable(sPAPRTCETable *tcet,
> >> tcet->nb_table = nb_table;
> >> tcet->table = spapr_tce_alloc_table(tcet->liobn,
> >> tcet->page_shift,
> >> + tcet->bus_offset,
> >> tcet->nb_table,
> >> &tcet->fd,
> >> tcet->need_vfio);
> >> diff --git a/target-ppc/kvm.c b/target-ppc/kvm.c
> >> index 9c4834c4fc..6e91a4d8bb 100644
> >> --- a/target-ppc/kvm.c
> >> +++ b/target-ppc/kvm.c
> >> @@ -71,6 +71,7 @@ static int cap_booke_sregs;
> >> static int cap_ppc_smt;
> >> static int cap_ppc_rma;
> >> static int cap_spapr_tce;
> >> +static int cap_spapr_tce_64;
> >> static int cap_spapr_multitce;
> >> static int cap_spapr_vfio;
> >> static int cap_hior;
> >> @@ -123,6 +124,7 @@ int kvm_arch_init(MachineState *ms, KVMState *s)
> >> cap_ppc_smt = kvm_check_extension(s, KVM_CAP_PPC_SMT);
> >> cap_ppc_rma = kvm_check_extension(s, KVM_CAP_PPC_RMA);
> >> cap_spapr_tce = kvm_check_extension(s, KVM_CAP_SPAPR_TCE);
> >> + cap_spapr_tce_64 = kvm_check_extension(s, KVM_CAP_SPAPR_TCE_64);
> >> cap_spapr_multitce = kvm_check_extension(s, KVM_CAP_SPAPR_MULTITCE);
> >> cap_spapr_vfio = false;
> >> cap_one_reg = kvm_check_extension(s, KVM_CAP_ONE_REG);
> >> @@ -2201,13 +2203,10 @@ bool kvmppc_spapr_use_multitce(void)
> >> return cap_spapr_multitce;
> >> }
> >>
> >> -void *kvmppc_create_spapr_tce(uint32_t liobn, uint32_t window_size, int *pfd,
> >> - bool need_vfio)
> >> +void *kvmppc_create_spapr_tce(uint32_t liobn, uint32_t page_shift,
> >> + uint64_t bus_offset, uint32_t nb_table,
> >> + int *pfd, bool need_vfio)
> >> {
> >> - struct kvm_create_spapr_tce args = {
> >> - .liobn = liobn,
> >> - .window_size = window_size,
> >> - };
> >> long len;
> >> int fd;
> >> void *table;
> >> @@ -2220,14 +2219,41 @@ void *kvmppc_create_spapr_tce(uint32_t liobn, uint32_t window_size, int *pfd,
> >> return NULL;
> >> }
> >>
> >> - fd = kvm_vm_ioctl(kvm_state, KVM_CREATE_SPAPR_TCE, &args);
> >> - if (fd < 0) {
> >> - fprintf(stderr, "KVM: Failed to create TCE table for liobn 0x%x\n",
> >> - liobn);
> >> + if (cap_spapr_tce_64) {
> >> + struct kvm_create_spapr_tce_64 args = {
> >> + .liobn = liobn,
> >> + .page_shift = page_shift,
> >> + .offset = bus_offset >> page_shift,
> >> + .size = nb_table,
> >> + .flags = 0
> >> + };
> >> + fd = kvm_vm_ioctl(kvm_state, KVM_CREATE_SPAPR_TCE_64, &args);
> >> + if (fd < 0) {
> >> + fprintf(stderr,
> >> + "KVM: Failed to create TCE64 table for liobn 0x%x\n",
> >> + liobn);
> >> + return NULL;
> >> + }
> >> + } else if (cap_spapr_tce) {
> >> + uint64_t window_size = (uint64_t) nb_table << page_shift;
> >> + struct kvm_create_spapr_tce args = {
> >> + .liobn = liobn,
> >> + .window_size = window_size,
> >> + };
> >> + if ((window_size != args.window_size) || bus_offset) {
> >> + return NULL;
> >> + }
> >> + fd = kvm_vm_ioctl(kvm_state, KVM_CREATE_SPAPR_TCE, &args);
> >> + if (fd < 0) {
> >> + fprintf(stderr, "KVM: Failed to create TCE table for liobn 0x%x\n",
> >> + liobn);
> >> + return NULL;
> >> + }
> >> + } else {
> >> return NULL;
> >> }
> >>
> >> - len = (window_size / SPAPR_TCE_PAGE_SIZE) * sizeof(uint64_t);
> >> + len = nb_table * sizeof(uint64_t);
> >> /* FIXME: round this up to page size */
> >>
> >> table = mmap(NULL, len, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0);
> >
>
>
--
David Gibson | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_
| _way_ _around_!
http://www.ozlabs.org/~dgibson
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]
next prev parent reply other threads:[~2017-01-09 2:54 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-12-22 1:13 [Qemu-devel] [PATCH qemu] target-ppc: kvm: make use of KVM_CREATE_SPAPR_TCE_64 Alexey Kardashevskiy
2017-01-03 2:26 ` David Gibson
2017-01-09 2:38 ` Alexey Kardashevskiy
2017-01-09 2:53 ` David Gibson [this message]
2017-03-02 2:40 ` Alexey Kardashevskiy
2017-03-08 4:48 ` Alexey Kardashevskiy
2017-03-10 0:32 ` David Gibson
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=20170109025349.GF12515@umbus.fritz.box \
--to=david@gibson.dropbear.id.au \
--cc=aik@ozlabs.ru \
--cc=qemu-devel@nongnu.org \
--cc=qemu-ppc@nongnu.org \
/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).