[PATCH 1/4] x86: Separate out the EPT logs from general VM logs

[hcha...@xvisor-x86.org]

From: Himanshu Chauhan <hcha...@xvisor-x86.org>

EPT logs in debug mode can be very noise. Separate them out.
Slowly, we will move every module to its own logging mechanism
so that they can be individually controlled.

Signed-off-by: Himanshu Chauhan <hcha...@xvisor-x86.org>
---
arch/x86/cpu/common/include/vm/ept.h | 30 +++++++++
arch/x86/cpu/common/vm/vtx/ept.c | 96 ++++++++++++++--------------
2 files changed, 79 insertions(+), 47 deletions(-)

diff --git a/arch/x86/cpu/common/include/vm/ept.h b/arch/x86/cpu/common/include/vm/ept.h
index 0d20fbc6..b54a65fe 100644
--- a/arch/x86/cpu/common/include/vm/ept.h
+++ b/arch/x86/cpu/common/include/vm/ept.h
@@ -189,4 +189,34 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
physical_addr_t gphys, physical_addr_t hphys,
size_t pg_size, u32 pg_prot);

+#define ENABLE_EPT_LOG 0
+
+enum {
+ EPT_LOG_LVL_ERR,
+ EPT_LOG_LVL_INFO,
+ EPT_LOG_LVL_DEBUG,
+ EPT_LOG_LVL_VERBOSE
+};
+extern int ept_default_log_lvl;
+
+#if ENABLE_EPT_LOG
+#define EPT_LOG(lvl, fmt, args...) \
+ do { \
+ if (EPT_LOG_##lvl <= ept_default_log_lvl) { \
+ vmm_printf(fmt, ##args); \
+ } \
+ }while(0);
+
+#define EPT_LOG_FD(lvl, fmt, args...) \
+ do { \
+ if (EPT_LOG_##lvl <= ept_default_log_lvl) { \
+ vmm_printf("(%s:%d) " fmt, __func__, \
+ __LINE__, ##args); \
+ } \
+ }while(0);
+#else
+#define EPT_LOG(lvl, fmt, args...)
+#define EPT_LOG_FD(lvl, fmt, args...)
+#endif
+
#endif /* __EPT_H */
diff --git a/arch/x86/cpu/common/vm/vtx/ept.c b/arch/x86/cpu/common/vm/vtx/ept.c
index ef7e42c1..dc412177 100644
--- a/arch/x86/cpu/common/vm/vtx/ept.c
+++ b/arch/x86/cpu/common/vm/vtx/ept.c
@@ -36,6 +36,8 @@
#include <vmm_main.h>
#include <vm/ept.h>

+int ept_default_log_lvl = EPT_LOG_LVL_DEBUG;
+
#define TRACE_EPT 1

#if TRACE_EPT
@@ -125,19 +127,19 @@ invalidate_ept (int type, struct invept_desc *desc)
/* most modern CPUs will have this */
if (unlikely(type == INVEPT_ALL_CONTEXT
&& !cpu_has_vmx_ept_invept_all_context)) {
- VM_LOG(LVL_DEBUG, "EPT all context flush not supported\n");
+ EPT_LOG(LVL_DEBUG, "EPT all context flush not supported\n");
return;
}
if (unlikely(type == INVEPT_SINGLE_CONTEXT
&& !cpu_has_vmx_ept_invept_single_context)) {
- VM_LOG(LVL_DEBUG, "EPT single context flush not supported\n");
+ EPT_LOG(LVL_DEBUG, "EPT single context flush not supported\n");
return;
}
asm volatile("invept (%0), %1\n\t"
::"D"(type), "S"(desc)
:"memory", "cc");
} else {
- VM_LOG(LVL_DEBUG, "INVEPT instruction is not supported by CPU\n");
+ EPT_LOG(LVL_DEBUG, "INVEPT instruction is not supported by CPU\n");
}
}

@@ -203,7 +205,7 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
int rc = 0;
struct invept_desc id;

- VM_LOG(LVL_DEBUG, "pml4: 0x%"PRIx32" pdpt: 0x%"PRIx32" pd: 0x%"PRIx32" pt: 0x%"PRIx32"\n",
+ EPT_LOG(LVL_DEBUG, "pml4: 0x%"PRIx32" pdpt: 0x%"PRIx32" pd: 0x%"PRIx32" pt: 0x%"PRIx32"\n",
pml4_index, pdpt_index, pd_index, pt_index);

add_ept_trace_point(gphys, hphys, pml4_index, pdpt_index, pd_index, pt_index, pg_size, pg_prot);
@@ -216,29 +218,29 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
pml4e->val |= pg_prot;
virt = get_free_page_for_pagemap(context, &phys);
if (!virt) {
- VM_LOG(LVL_ERR, "System is out of guest page table memory\n");
+ EPT_LOG(LVL_ERR, "System is out of guest page table memory\n");
rc = VMM_ENOMEM;
goto _done;
}
- VM_LOG(LVL_DEBUG, "New PDPT Page at 0x%"PRIx64" (Phys: 0x%"PRIx64") for PML4 Index %d.\n",
+ EPT_LOG(LVL_DEBUG, "New PDPT Page at 0x%"PRIx64" (Phys: 0x%"PRIx64") for PML4 Index %d.\n",
virt, phys, pml4_index);
memset((void *)virt, 0, PAGE_SIZE);
pml4e->bits.pdpt_base = EPT_PHYS_4KB_PFN(phys);
} else {
if (vmm_host_pa2va(e_phys, &virt) != VMM_OK) {
- VM_LOG(LVL_ERR, "Couldn't map PDPTE physical 0x%"PRIx64" to virtual\n",
+ EPT_LOG(LVL_ERR, "Couldn't map PDPTE physical 0x%"PRIx64" to virtual\n",
e_phys);
rc = VMM_ENOENT;
goto _done;
}
- VM_LOG(LVL_DEBUG, "Found PDPT Page at 0x%"PRIx64" (phys: 0x%"PRIx64") for PML4 Index: %d\n",
+ EPT_LOG(LVL_DEBUG, "Found PDPT Page at 0x%"PRIx64" (phys: 0x%"PRIx64") for PML4 Index: %d\n",
virt, e_phys, pml4_index);
}
- VM_LOG(LVL_DEBUG, "%s: PML4E: 0x%"PRIx64"\n", __func__, pml4e->val);
+ EPT_LOG(LVL_DEBUG, "%s: PML4E: 0x%"PRIx64"\n", __func__, pml4e->val);

phys = e_phys = e_pg_prot = 0;
pdpte = (ept_pdpte_t *)(&((u64 *)virt)[pdpt_index]);
- VM_LOG(LVL_DEBUG, "%s: PDPTE: 0x%"PRIx64" (PDPT Index: %d)\n", __func__, pdpte->val, pdpt_index);
+ EPT_LOG(LVL_DEBUG, "%s: PDPTE: 0x%"PRIx64" (PDPT Index: %d)\n", __func__, pdpte->val, pdpt_index);
decode_ept_entry(EPT_LEVEL_PDPTE, (void *)pdpte, &e_phys, &e_pg_prot);

/*
@@ -247,12 +249,12 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
* up the other one.
*/
if (pdpte->pe.is_page) {
- VM_LOG(LVL_DEBUG, "PDPTE is page\n");
+ EPT_LOG(LVL_DEBUG, "PDPTE is page\n");
/* this is marked as 1GB page and new mapping wants otherwise
* then its a problem. Caller didn't free this mapping prior
* to calling this function */
if (pg_size != EPT_PAGE_SIZE_1G) {
- VM_LOG(LVL_DEBUG, "New page size is not 1G (0x%"PRIx64"). Delete existing entry first.\n", pg_size);
+ EPT_LOG(LVL_DEBUG, "New page size is not 1G (0x%"PRIx64"). Delete existing entry first.\n", pg_size);
rc = VMM_EBUSY;
goto _done;
}
@@ -277,7 +279,7 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
}

if (pg_size == EPT_PAGE_SIZE_1G) {
- VM_LOG(LVL_DEBUG, "Creating map of 1G page at pdpt index: %d\n", pdpt_index);
+ EPT_LOG(LVL_DEBUG, "Creating map of 1G page at pdpt index: %d\n", pdpt_index);
pdpte->val = 0;
pdpte->val &= EPT_PROT_MASK;
pdpte->val |= pg_prot;
@@ -285,18 +287,18 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
pdpte->pe.mt = 6; /* write-back memory type */
pdpte->pe.ign_pat = 1; /* ignore PAT type */
pdpte->pe.is_page = 1;
- VM_LOG(LVL_DEBUG, "New PDPT Entry: 0x%"PRIx64"\n", pdpte->val);
+ EPT_LOG(LVL_DEBUG, "New PDPT Entry: 0x%"PRIx64"\n", pdpte->val);
rc = VMM_OK;
/* new entry. Invalidate EPT */
goto _invalidate_ept;
} else { /* not a 1G page */
- VM_LOG(LVL_DEBUG, "PDPTE doesn't point to 1G page. Looking for PDE\n");
+ EPT_LOG(LVL_DEBUG, "PDPTE doesn't point to 1G page. Looking for PDE\n");
if (!e_pg_prot) { /* if the page is not currently set */
- VM_LOG(LVL_DEBUG, "PDE page protection not set. Creating new one\n");
+ EPT_LOG(LVL_DEBUG, "PDE page protection not set. Creating new one\n");
virt = get_free_page_for_pagemap(context, &phys);
/* allocate a new PDPTE page */
if (!virt) {
- VM_LOG(LVL_ERR, "System is out of guest page table memory\n");
+ EPT_LOG(LVL_ERR, "System is out of guest page table memory\n");
rc = VMM_ENOMEM;
goto _done;
}
@@ -305,53 +307,53 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
pdpte->te.pd_base = EPT_PHYS_4KB_PFN(phys);
pdpte->val &= EPT_PROT_MASK;
pdpte->val |= pg_prot;
- VM_LOG(LVL_DEBUG, "New PD Page at 0x%"PRIx64" (Phys: 0x%"PRIx64")\n", virt, phys);
+ EPT_LOG(LVL_DEBUG, "New PD Page at 0x%"PRIx64" (Phys: 0x%"PRIx64")\n", virt, phys);
} else { /* page is already allocated, a mapping in locality exists */
if (vmm_host_pa2va(e_phys, &virt) != VMM_OK) {
- VM_LOG(LVL_ERR, "Couldn't map PDE physical 0x%"PRIx64" to virtual\n",
+ EPT_LOG(LVL_ERR, "Couldn't map PDE physical 0x%"PRIx64" to virtual\n",
e_phys);
rc = VMM_ENOENT;
goto _done;
}
- VM_LOG(LVL_DEBUG, "Found PDE at virtual address 0x%"PRIx64"\n", virt);
+ EPT_LOG(LVL_DEBUG, "Found PDE at virtual address 0x%"PRIx64"\n", virt);
}
}
- VM_LOG(LVL_DEBUG, "%s: PDPTE: 0x%"PRIx64"\n", __func__, pdpte->val);
+ EPT_LOG(LVL_DEBUG, "%s: PDPTE: 0x%"PRIx64"\n", __func__, pdpte->val);

phys = e_phys = e_pg_prot = 0;
pde = (ept_pde_t *)(&((u64 *)virt)[pd_index]);
- VM_LOG(LVL_DEBUG, "PDPTE Entry at index %d = 0x%"PRIx64"\n", pd_index, pde->val);
+ EPT_LOG(LVL_DEBUG, "PDPTE Entry at index %d = 0x%"PRIx64"\n", pd_index, pde->val);
decode_ept_entry(EPT_LEVEL_PDE, (void *)pde, &e_phys, &e_pg_prot);

if (pde->pe.is_page) {
- VM_LOG(LVL_DEBUG, "PDE is a 2MB Page!\n");
+ EPT_LOG(LVL_DEBUG, "PDE is a 2MB Page!\n");
/* this is marked as 1GB page and new mapping wants otherwise
* then its a problem. Caller didn't free this mapping prior
* to calling this function */
if (pg_size != EPT_PAGE_SIZE_2M) {
- VM_LOG(LVL_DEBUG, "New page is not 2M. Delete previous entry first.\n");
+ EPT_LOG(LVL_DEBUG, "New page is not 2M. Delete previous entry first.\n");
rc = VMM_EBUSY;
goto _done;
}

/* caller is trying to create same mapping? */
if (e_phys == hphys) {
- VM_LOG(LVL_DEBUG, "Found same physical addres at pd index: %d\n", pd_index);
+ EPT_LOG(LVL_DEBUG, "Found same physical addres at pd index: %d\n", pd_index);
if (pg_prot != e_pg_prot) {
- VM_LOG(LVL_DEBUG, "PG prot are not same. Old: 0x%"PRIx32" New: 0x%"PRIx32"\n",
+ EPT_LOG(LVL_DEBUG, "PG prot are not same. Old: 0x%"PRIx32" New: 0x%"PRIx32"\n",
e_pg_prot, pg_prot);
pde->val |= pg_prot;
rc = VMM_OK;
/* pgprot changed, invalidate ept */
goto _invalidate_ept;
} else {
- VM_LOG(LVL_DEBUG, "No change in page table entry.\n");
+ EPT_LOG(LVL_DEBUG, "No change in page table entry.\n");
/* no change, same as existing mapping */
rc = VMM_OK;
goto _done;
}
} else {
- VM_LOG(LVL_DEBUG, "pd index %d is busy. Val: 0x%"PRIx64"\n", pd_index, pde->val);
+ EPT_LOG(LVL_DEBUG, "pd index %d is busy. Val: 0x%"PRIx64"\n", pd_index, pde->val);
/* existing physical is not same as new one. flag as error.
* caller should have unmapped this mapping first */
rc = VMM_EBUSY;
@@ -361,7 +363,7 @@ int ept_create_pte_map(struct vcpu_hw_context *context,

/* not a 2MB page, is caller trying to create a 2MB page? */
if (pg_size == EPT_PAGE_SIZE_2M) {
- VM_LOG(LVL_DEBUG, "Ask is to create 2MB page\n");
+ EPT_LOG(LVL_DEBUG, "Ask is to create 2MB page\n");
pdpte->val = 0;
pdpte->val &= EPT_PROT_MASK;
pdpte->val |= pg_prot;
@@ -369,17 +371,17 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
pdpte->pe.mt = 6; /* write-back memory type */
pdpte->pe.ign_pat = 1; /* ignore PAT type */
pdpte->pe.is_page = 1;
- VM_LOG(LVL_DEBUG, "New 2MB page. PDE Value: 0x%"PRIx64" at index: %d.\n", pdpte->val, pd_index);
+ EPT_LOG(LVL_DEBUG, "New 2MB page. PDE Value: 0x%"PRIx64" at index: %d.\n", pdpte->val, pd_index);
rc = VMM_OK;
goto _invalidate_ept;
} else {
/* Ok. So this is PDE. Lets find PTE now. */
if (!e_pg_prot) { /* page for PTE is not currently set */
- VM_LOG(LVL_DEBUG, "Page protection bits not set in PTE page. Creating new one.\n");
+ EPT_LOG(LVL_DEBUG, "Page protection bits not set in PTE page. Creating new one.\n");
virt = get_free_page_for_pagemap(context, &phys);
/* allocate a new PTE page */
if (!virt) {
- VM_LOG(LVL_ERR, "System is out of guest page table memory\n");
+ EPT_LOG(LVL_ERR, "System is out of guest page table memory\n");
rc = VMM_ENOMEM;
goto _done;
}
@@ -388,30 +390,30 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
pde->te.pt_base = EPT_PHYS_4KB_PFN(phys);
pde->val &= EPT_PROT_MASK;
pde->val |= pg_prot;
- VM_LOG(LVL_DEBUG, "New PT page at 0x%"PRIx64" (Phys: 0x%"PRIx64")\n",
+ EPT_LOG(LVL_DEBUG, "New PT page at 0x%"PRIx64" (Phys: 0x%"PRIx64")\n",
virt, phys);
} else { /* page is already allocated, a mapping in locality exists */
if (vmm_host_pa2va(e_phys, &virt) != VMM_OK) {
- VM_LOG(LVL_ERR, "Couldn't map PDE physical 0x%"PRIx64" to virtual\n",
+ EPT_LOG(LVL_ERR, "Couldn't map PDE physical 0x%"PRIx64" to virtual\n",
e_phys);
rc = VMM_ENOENT;
goto _done;
}
- VM_LOG(LVL_DEBUG, "Found PT at virt 0x%"PRIx64"\n", virt);
+ EPT_LOG(LVL_DEBUG, "Found PT at virt 0x%"PRIx64"\n", virt);
}
}
- VM_LOG(LVL_DEBUG, "%s: PDE: 0x%"PRIx64"\n", __func__, pde->val);
+ EPT_LOG(LVL_DEBUG, "%s: PDE: 0x%"PRIx64"\n", __func__, pde->val);

e_phys = e_pg_prot = 0;
pte = (ept_pte_t *)(&((u64 *)virt)[pt_index]);
- VM_LOG(LVL_DEBUG, "PT Entry 0x%"PRIx64" at index: %d\n", pte->val, pt_index);
+ EPT_LOG(LVL_DEBUG, "PT Entry 0x%"PRIx64" at index: %d\n", pte->val, pt_index);
decode_ept_entry(EPT_LEVEL_PTE, (void *)pte, &e_phys, &e_pg_prot);
if (e_pg_prot) { /* mapping exists */
- VM_LOG(LVL_DEBUG, "Page mapping exists: current pgprot: 0x%"PRIx32"\n", e_pg_prot);
+ EPT_LOG(LVL_DEBUG, "Page mapping exists: current pgprot: 0x%"PRIx32"\n", e_pg_prot);
if (e_phys == hphys) {
- VM_LOG(LVL_DEBUG, "Existing physical and asked are same. (e_phys: 0x%"PRIx64" h_phys: 0x%"PRIx64")\n", e_phys, hphys);
+ EPT_LOG(LVL_DEBUG, "Existing physical and asked are same. (e_phys: 0x%"PRIx64" h_phys: 0x%"PRIx64")\n", e_phys, hphys);
if (e_pg_prot == pg_prot) { /* same mapping */
- VM_LOG(LVL_DEBUG, "Same PG prot: old: 0x%"PRIx32" new: 0x%"PRIx32"\n", e_pg_prot, pg_prot);
+ EPT_LOG(LVL_DEBUG, "Same PG prot: old: 0x%"PRIx32" new: 0x%"PRIx32"\n", e_pg_prot, pg_prot);
rc = VMM_OK;
goto _done; /* no change */
}
@@ -420,25 +422,25 @@ int ept_create_pte_map(struct vcpu_hw_context *context,
pte->val |= pg_prot;
goto _invalidate_ept;
} else {
- VM_LOG(LVL_DEBUG, "Existing PTE entry found at index: %d but with phys: 0x%"PRIx64" (new: 0x%"PRIx64")\n",
+ EPT_LOG(LVL_DEBUG, "Existing PTE entry found at index: %d but with phys: 0x%"PRIx64" (new: 0x%"PRIx64")\n",
pt_index, e_phys, hphys);
rc = VMM_EBUSY;
goto _done;
}
} else {
- VM_LOG(LVL_DEBUG, "No page protection bits set in PTE. Creating new one\n");
+ EPT_LOG(LVL_DEBUG, "No page protection bits set in PTE. Creating new one\n");
pte->val = 0;
pte->val &= EPT_PROT_MASK;
pte->val |= pg_prot;
pte->pe.mt = 6;
pte->pe.phys = EPT_PHYS_4KB_PFN(hphys);
rc = VMM_OK;
- VM_LOG(LVL_DEBUG, "%s: PTE: 0x%"PRIx64" at index %d\n", __func__, pte->val, pt_index);
+ EPT_LOG(LVL_DEBUG, "%s: PTE: 0x%"PRIx64" at index %d\n", __func__, pte->val, pt_index);
goto _invalidate_ept;
}

_invalidate_ept:
- VM_LOG(LVL_DEBUG, "Invalidating EPT\n");
+ EPT_LOG(LVL_DEBUG, "Invalidating EPT\n");

id.eptp = context->eptp;
invalidate_ept(INVEPT_SINGLE_CONTEXT, &id);
@@ -452,11 +454,11 @@ int setup_ept(struct vcpu_hw_context *context)
eptp_t *eptp = (eptp_t *)&context->eptp;
virtual_addr_t pml4 = get_free_page_for_pagemap(context, &pml4_phys);

- VM_LOG(LVL_DEBUG, "%s: PML4 vaddr: 0x%016lx paddr: 0x%016lx\n",
+ EPT_LOG(LVL_DEBUG, "%s: PML4 vaddr: 0x%016lx paddr: 0x%016lx\n",
__func__, pml4, pml4_phys);

if (!pml4) {
- VM_LOG(LVL_ERR, "%s: Failed to allocate EPT page\n", __func__);
+ EPT_LOG(LVL_ERR, "%s: Failed to allocate EPT page\n", __func__);
return VMM_ENOMEM;

}
@@ -476,7 +478,7 @@ int setup_ept(struct vcpu_hw_context *context)
eptp->bits.en_ad = 0;
eptp->bits.pml4 = EPT_PHYS_4KB_PFN(pml4_phys);

- VM_LOG(LVL_DEBUG, "%s: EPTP: 0x%16lx (0x%16lx)\n", __func__, eptp->val, context->eptp);
+ EPT_LOG(LVL_DEBUG, "%s: EPTP: 0x%16lx (0x%16lx)\n", __func__, eptp->val, context->eptp);

context->n_cr3 = pml4;

--
2.25.1

[hcha...@xvisor-x86.org]

From: Himanshu Chauhan <hcha...@xvisor-x86.org>

Signed-off-by: Himanshu Chauhan <hcha...@xvisor-x86.org>
---

emulators/include/emu/pci/pci_emu_core.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/emulators/include/emu/pci/pci_emu_core.h b/emulators/include/emu/pci/pci_emu_core.h
index 5656919f..0402a4c2 100644
--- a/emulators/include/emu/pci/pci_emu_core.h
+++ b/emulators/include/emu/pci/pci_emu_core.h
@@ -89,9 +89,9 @@ struct pci_conf_header {
u16 command;
u16 status;
u8 revision;
- u8 class;
- u8 sub_class;
u8 prog_if;
+ u8 sub_class;
+ u8 class;
u8 cache_line_sz;
u8 latency_timer;
u8 header_type;
--
2.25.1

[hcha...@xvisor-x86.org]

From: Himanshu Chauhan <hcha...@xvisor-x86.org>

There is a bug in virtio-pci emulation where the type of the PCI device
is not reported correctly. Seabios mistakes it with memory-mapped device
when it is an IO mapped devide. This causes seabios to reservce huge
physical address space for this device. The 32-bit address space runs out.
Right now, removing virtio-pci emulation.

Enable LAPIC emulation.

Signed-off-by: Himanshu Chauhan <hcha...@xvisor-x86.org>
---

tests/x86/intel-guest.dts | 26 ++------------------------
1 file changed, 2 insertions(+), 24 deletions(-)

diff --git a/tests/x86/intel-guest.dts b/tests/x86/intel-guest.dts
index 4de9382b..ddae01b8 100644
--- a/tests/x86/intel-guest.dts
+++ b/tests/x86/intel-guest.dts
@@ -107,28 +107,6 @@

pci_bus0 {
devices {
- virtio-blk0 {
- device_type = "virtio";
- compatible = "virtio,pci";
- manifest_type = "virtual";
- address_type = "io";
- device_id = <2>;
-
- bars {
- bar0@virtio-blk0 {
- manifest_type = "virtual";
- address_type = "io";
- device_type = "virtio";
- compatible = "virtio,pci,bar";
- barnum = <0>;
- virtio_type = <2>;
- guest_physical_addr = <0x610>;
- physical_size = <0x100>;
- blkdev = "hda0";
- interrupts = <55>;
- };
- };
- };
};
};
};
@@ -170,7 +148,7 @@
parent_irq = <2>;
};

-/*
+
lapic {
device_type = "pic";
compatible = "lapic";
@@ -181,7 +159,7 @@
base_irq = <16>;
num_irq = <240>;
};
-*/
+
hpet {
device_type = "hpet";
compatible = "hpet";
--
2.25.1

[hcha...@xvisor-x86.org]

From: Himanshu Chauhan <hcha...@xvisor-x86.org>

Class code was missing in i440FX PCI config space. Seabios
had trouble understanding that its a bridge. Add class code
so that emulation of bridge works well.

Signed-off-by: Himanshu Chauhan <hcha...@xvisor-x86.org>
---

emulators/pci/host/i440fx.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/emulators/pci/host/i440fx.c b/emulators/pci/host/i440fx.c
index 27035456..bf1d18bd 100644
--- a/emulators/pci/host/i440fx.c
+++ b/emulators/pci/host/i440fx.c
@@ -331,6 +331,8 @@ static int i440fx_emulator_probe(struct vmm_guest *guest,
INIT_SPIN_LOCK(&class->lock);
class->conf_header.vendor_id = PCI_VENDOR_ID_INTEL;
class->conf_header.device_id = PCI_DEVICE_ID_INTEL_82441;
+ class->conf_header.class = PCI_BASE_CLASS_BRIDGE;
+ class->conf_header.revision = 2;
class->config_read = i440fx_config_read;
class->config_write = i440fx_config_write;

--
2.25.1

[Anup Patel]

On Mon, May 23, 2022 at 11:46 PM <hcha...@xvisor-x86.org> wrote:
>
> From: Himanshu Chauhan <hcha...@xvisor-x86.org>
>
> EPT logs in debug mode can be very noise. Separate them out.
> Slowly, we will move every module to its own logging mechanism
> so that they can be individually controlled.
>
> Signed-off-by: Himanshu Chauhan <hcha...@xvisor-x86.org>

Looks good to me.

Reviewed-by: Anup Patel <an...@brainfault.org>

Thanks,
Anup

> --
> You received this message because you are subscribed to the Google Groups "Xvisor Development" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to xvisor-devel...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/xvisor-devel/20220523181638.1227684-1-hchauhan%40xvisor-x86.org.

[Anup Patel]

On Mon, May 23, 2022 at 11:46 PM <hcha...@xvisor-x86.org> wrote:
>

> From: Himanshu Chauhan <hcha...@xvisor-x86.org>
>
> Signed-off-by: Himanshu Chauhan <hcha...@xvisor-x86.org>

Looks good to me.

Reviewed-by: Anup Patel <an...@brainfault.org>

Thanks,
Anup

> ---
> emulators/include/emu/pci/pci_emu_core.h | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/emulators/include/emu/pci/pci_emu_core.h b/emulators/include/emu/pci/pci_emu_core.h
> index 5656919f..0402a4c2 100644
> --- a/emulators/include/emu/pci/pci_emu_core.h
> +++ b/emulators/include/emu/pci/pci_emu_core.h
> @@ -89,9 +89,9 @@ struct pci_conf_header {
> u16 command;
> u16 status;
> u8 revision;
> - u8 class;
> - u8 sub_class;
> u8 prog_if;
> + u8 sub_class;
> + u8 class;
> u8 cache_line_sz;
> u8 latency_timer;
> u8 header_type;
> --
> 2.25.1
>

> --
> You received this message because you are subscribed to the Google Groups "Xvisor Development" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to xvisor-devel...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/xvisor-devel/20220523181638.1227684-2-hchauhan%40xvisor-x86.org.

[Anup Patel]

On Mon, May 23, 2022 at 11:46 PM <hcha...@xvisor-x86.org> wrote:
>

> From: Himanshu Chauhan <hcha...@xvisor-x86.org>
>
> Class code was missing in i440FX PCI config space. Seabios
> had trouble understanding that its a bridge. Add class code
> so that emulation of bridge works well.
>
> Signed-off-by: Himanshu Chauhan <hcha...@xvisor-x86.org>

Looks good to me.

Reviewed-by: Anup Patel <an...@brainfault.org>

Thanks,
Anup

> ---
> emulators/pci/host/i440fx.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/emulators/pci/host/i440fx.c b/emulators/pci/host/i440fx.c
> index 27035456..bf1d18bd 100644
> --- a/emulators/pci/host/i440fx.c
> +++ b/emulators/pci/host/i440fx.c
> @@ -331,6 +331,8 @@ static int i440fx_emulator_probe(struct vmm_guest *guest,
> INIT_SPIN_LOCK(&class->lock);
> class->conf_header.vendor_id = PCI_VENDOR_ID_INTEL;
> class->conf_header.device_id = PCI_DEVICE_ID_INTEL_82441;
> + class->conf_header.class = PCI_BASE_CLASS_BRIDGE;
> + class->conf_header.revision = 2;
> class->config_read = i440fx_config_read;
> class->config_write = i440fx_config_write;
>
> --
> 2.25.1
>

> --
> You received this message because you are subscribed to the Google Groups "Xvisor Development" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to xvisor-devel...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/xvisor-devel/20220523181638.1227684-3-hchauhan%40xvisor-x86.org.

[Anup Patel]

On Mon, May 23, 2022 at 11:46 PM <hcha...@xvisor-x86.org> wrote:
>

> From: Himanshu Chauhan <hcha...@xvisor-x86.org>
>
> There is a bug in virtio-pci emulation where the type of the PCI device
> is not reported correctly. Seabios mistakes it with memory-mapped device
> when it is an IO mapped devide. This causes seabios to reservce huge
> physical address space for this device. The 32-bit address space runs out.
> Right now, removing virtio-pci emulation.
>
> Enable LAPIC emulation.
>
> Signed-off-by: Himanshu Chauhan <hcha...@xvisor-x86.org>

Looks good to me.

Reviewed-by: Anup Patel <an...@brainfault.org>

Thanks,
Anup

> --
> You received this message because you are subscribed to the Google Groups "Xvisor Development" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to xvisor-devel...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/xvisor-devel/20220523181638.1227684-4-hchauhan%40xvisor-x86.org.