[xen staging] tools/pvh: set coherent MTRR state for all vCPUs

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[xen staging] tools/pvh: set coherent MTRR state for all vCPUs

patchbot
commit 92666fdd6e0afab989b2d89759d9b43f2c645ae7
Author:     Roger Pau Monne <[hidden email]>
AuthorDate: Wed Oct 10 16:39:35 2018 +0200
Commit:     Wei Liu <[hidden email]>
CommitDate: Thu Oct 11 09:42:38 2018 +0100

    tools/pvh: set coherent MTRR state for all vCPUs
   
    Instead of just doing it for the BSP. This requires storing the
    maximum number of possible vCPUs in xc_dom_image.
   
    This has been a latent bug so far because PVH doesn't yet support
    pci-passthrough, so the effective memory cache attribute is forced to
    WB by the hypervisor. Note also that even without this in place vCPU#0
    is preferred in certain scenarios in order to calculate the memory
    cache attributes.
   
    Reported-by: Andrew Cooper <[hidden email]>
    Signed-off-by: Roger Pau Monné <[hidden email]>
    Acked-by: Wei Liu <[hidden email]>
---
 tools/libxc/include/xc_dom.h |  3 ++
 tools/libxc/xc_dom_x86.c     | 71 ++++++++++++++++++++++++++++++--------------
 tools/libxl/libxl_dom.c      |  2 ++
 3 files changed, 54 insertions(+), 22 deletions(-)

diff --git a/tools/libxc/include/xc_dom.h b/tools/libxc/include/xc_dom.h
index 0b5a632d3c..5900bbe8fa 100644
--- a/tools/libxc/include/xc_dom.h
+++ b/tools/libxc/include/xc_dom.h
@@ -230,6 +230,9 @@ struct xc_dom_image {
 #endif
 
     xen_pfn_t vuart_gfn;
+
+    /* Number of vCPUs */
+    unsigned int max_vcpus;
 };
 
 /* --- pluggable kernel loader ------------------------------------- */
diff --git a/tools/libxc/xc_dom_x86.c b/tools/libxc/xc_dom_x86.c
index d77f2d6f62..77a4c6ccd0 100644
--- a/tools/libxc/xc_dom_x86.c
+++ b/tools/libxc/xc_dom_x86.c
@@ -955,17 +955,16 @@ static int vcpu_hvm(struct xc_dom_image *dom)
         HVM_SAVE_TYPE(HEADER) header;
         struct hvm_save_descriptor cpu_d;
         HVM_SAVE_TYPE(CPU) cpu;
-        struct hvm_save_descriptor mtrr_d;
-        HVM_SAVE_TYPE(MTRR) mtrr;
         struct hvm_save_descriptor end_d;
         HVM_SAVE_TYPE(END) end;
     } bsp_ctx;
-    const HVM_SAVE_TYPE(MTRR) *mtrr_record;
     uint8_t *full_ctx = NULL;
     int rc;
 
     DOMPRINTF_CALLED(dom->xch);
 
+    assert(dom->max_vcpus);
+
     /*
      * Get the full HVM context in order to have the header, it is not
      * possible to get the header with getcontext_partial, and crafting one
@@ -1034,35 +1033,63 @@ static int vcpu_hvm(struct xc_dom_image *dom)
     if ( dom->start_info_seg.pfn )
         bsp_ctx.cpu.rbx = dom->start_info_seg.pfn << PAGE_SHIFT;
 
-    /* Set the MTRR. */
-    bsp_ctx.mtrr_d.typecode = HVM_SAVE_CODE(MTRR);
-    bsp_ctx.mtrr_d.instance = 0;
-    bsp_ctx.mtrr_d.length = HVM_SAVE_LENGTH(MTRR);
+    /* Set the end descriptor. */
+    bsp_ctx.end_d.typecode = HVM_SAVE_CODE(END);
+    bsp_ctx.end_d.instance = 0;
+    bsp_ctx.end_d.length = HVM_SAVE_LENGTH(END);
 
-    mtrr_record = hvm_get_save_record(full_ctx, HVM_SAVE_CODE(MTRR), 0);
-    if ( !mtrr_record )
+    /* TODO: maybe this should be a firmware option instead? */
+    if ( !dom->device_model )
     {
-        xc_dom_panic(dom->xch, XC_INTERNAL_ERROR,
-                     "%s: unable to get MTRR save record", __func__);
-        goto out;
-    }
+        struct {
+            struct hvm_save_descriptor header_d;
+            HVM_SAVE_TYPE(HEADER) header;
+            struct hvm_save_descriptor mtrr_d;
+            HVM_SAVE_TYPE(MTRR) mtrr;
+            struct hvm_save_descriptor end_d;
+            HVM_SAVE_TYPE(END) end;
+        } mtrr = {
+            .header_d = bsp_ctx.header_d,
+            .header = bsp_ctx.header,
+            .mtrr_d.typecode = HVM_SAVE_CODE(MTRR),
+            .mtrr_d.length = HVM_SAVE_LENGTH(MTRR),
+            .end_d = bsp_ctx.end_d,
+            .end = bsp_ctx.end,
+        };
+        const HVM_SAVE_TYPE(MTRR) *mtrr_record =
+            hvm_get_save_record(full_ctx, HVM_SAVE_CODE(MTRR), 0);
+        unsigned int i;
+
+        if ( !mtrr_record )
+        {
+            xc_dom_panic(dom->xch, XC_INTERNAL_ERROR,
+                         "%s: unable to get MTRR save record", __func__);
+            goto out;
+        }
 
-    memcpy(&bsp_ctx.mtrr, mtrr_record, sizeof(bsp_ctx.mtrr));
+        memcpy(&mtrr.mtrr, mtrr_record, sizeof(mtrr.mtrr));
 
-    /* TODO: maybe this should be a firmware option instead? */
-    if ( !dom->device_model )
         /*
          * Enable MTRR, set default type to WB.
          * TODO: add MMIO areas as UC when passthrough is supported.
          */
-        bsp_ctx.mtrr.msr_mtrr_def_type = MTRR_TYPE_WRBACK |
-                                         MTRR_DEF_TYPE_ENABLE;
+        mtrr.mtrr.msr_mtrr_def_type = MTRR_TYPE_WRBACK | MTRR_DEF_TYPE_ENABLE;
 
-    /* Set the end descriptor. */
-    bsp_ctx.end_d.typecode = HVM_SAVE_CODE(END);
-    bsp_ctx.end_d.instance = 0;
-    bsp_ctx.end_d.length = HVM_SAVE_LENGTH(END);
+        for ( i = 0; i < dom->max_vcpus; i++ )
+        {
+            mtrr.mtrr_d.instance = i;
+            rc = xc_domain_hvm_setcontext(dom->xch, dom->guest_domid,
+                                          (uint8_t *)&mtrr, sizeof(mtrr));
+            if ( rc != 0 )
+                xc_dom_panic(dom->xch, XC_INTERNAL_ERROR,
+                             "%s: SETHVMCONTEXT failed (rc=%d)", __func__, rc);
+        }
+    }
 
+    /*
+     * Loading the BSP context should be done in the last call to setcontext,
+     * since each setcontext call will put all vCPUs down.
+     */
     rc = xc_domain_hvm_setcontext(dom->xch, dom->guest_domid,
                                   (uint8_t *)&bsp_ctx, sizeof(bsp_ctx));
     if ( rc != 0 )
diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
index 8a8a32c699..c66f3893d7 100644
--- a/tools/libxl/libxl_dom.c
+++ b/tools/libxl/libxl_dom.c
@@ -803,6 +803,7 @@ int libxl__build_pv(libxl__gc *gc, uint32_t domid,
     dom->xenstore_evtchn = state->store_port;
     dom->xenstore_domid = state->store_domid;
     dom->claim_enabled = libxl_defbool_val(info->claim_mode);
+    dom->max_vcpus = info->max_vcpus;
 
     if (info->num_vnuma_nodes != 0) {
         unsigned int i;
@@ -1256,6 +1257,7 @@ int libxl__build_hvm(libxl__gc *gc, uint32_t domid,
     dom->mmio_start = mmio_start;
     dom->vga_hole_size = device_model ? LIBXL_VGA_HOLE_SIZE : 0;
     dom->device_model = device_model;
+    dom->max_vcpus = info->max_vcpus;
 
     rc = libxl__domain_device_construct_rdm(gc, d_config,
                                             info->u.hvm.rdm_mem_boundary_memkb*1024,
--
generated by git-patchbot for /home/xen/git/xen.git#staging


_______________________________________________
Xen-changelog mailing list
[hidden email]
https://lists.xenproject.org/xen-changelog