This has been used to boot a full Linux kernel environment
to multiuser.
Change-Id: I12c399aafd30cdf05208831ca30129753f027f2a
Signed-off-by: Ronald G. Minnich <[email protected]>
---
tests/vmm/Makefrag | 2 +-
tests/vmm/vmrunkernel.c | 192 +++++++++++++++++++++++++++++++++++-------------
2 files changed, 142 insertions(+), 52 deletions(-)
diff --git a/tests/vmm/Makefrag b/tests/vmm/Makefrag
index 4208f63..49eb20f 100644
--- a/tests/vmm/Makefrag
+++ b/tests/vmm/Makefrag
@@ -4,7 +4,7 @@ VMM_TESTS_CFLAGS += $(CFLAGS_TESTS)
ALL_VMM_TEST_FILES := $(wildcard $(VMM_TESTS_DIR)/*.c)
-VMM_TESTS_LDLIBS := $(TESTS_LDLIBS)
+VMM_TESTS_LDLIBS := $(TESTS_LDLIBS) -lelf
VMM_TESTS_SRCS := $(ALL_VMM_TEST_FILES)
diff --git a/tests/vmm/vmrunkernel.c b/tests/vmm/vmrunkernel.c
index ea2ac5a..5f2787f 100644
--- a/tests/vmm/vmrunkernel.c
+++ b/tests/vmm/vmrunkernel.c
@@ -6,7 +6,9 @@
#include <parlib/arch/arch.h>
#include <parlib/ros_debug.h>
#include <unistd.h>
+#include <gelf.h>
#include <errno.h>
+#include <libelf.h>
#include <dirent.h>
#include <stdlib.h>
#include <string.h>
@@ -36,6 +38,9 @@
#include <sys/eventfd.h>
#include <sys/uio.h>
+#define DBG(format, ...) if (debug)\
+ fprintf(stderr, format, __VA_ARGS__)
+
struct virtual_machine local_vm, *vm = &local_vm;
struct vmm_gpcore_init gpci;
@@ -292,13 +297,13 @@ static uint8_t acpi_tb_checksum(uint8_t *buffer, uint32_t
length)
{
uint8_t sum = 0;
uint8_t *end = buffer + length;
- fprintf(stderr, "tbchecksum %p for %d", buffer, length);
+ DBG("tbchecksum %p for %d", buffer, length);
while (buffer < end) {
- if (end - buffer < 2)
+ if (0 && end - buffer < 2)
fprintf(stderr, "%02x\n", sum);
sum = (uint8_t)(sum + *(buffer++));
}
- fprintf(stderr, " is %02x\n", sum);
+ DBG(" is %02x\n", sum);
return (sum);
}
@@ -307,11 +312,11 @@ static void gencsum(uint8_t *target, void *data, int len)
uint8_t csum;
// blast target to zero so it does not get counted
// (it might be in the struct we checksum) And, yes, it is, goodness.
- fprintf(stderr, "gencsum %p target %p source %d bytes\n", target, data,
len);
+ DBG("gencsum %p target %p source %d bytes\n", target, data, len);
*target = 0;
csum = acpi_tb_checksum((uint8_t *)data, len);
*target = ~csum + 1;
- fprintf(stderr, "Cmoputed is %02x\n", *target);
+ DBG("ACPI: genchecksum: Computed is %02x\n", *target);
}
static inline int test_and_set_bit(int nr, volatile unsigned long *addr)
@@ -324,15 +329,108 @@ static inline int test_and_set_bit(int nr, volatile
unsigned long *addr)
return oldbit;
}
-static void pir_dump()
+/* load_kernel loads an ELF file as a kernel. */
+uintptr_t
+load_kernel(char *filename, uintptr_t *kernstart, uintptr_t *kernend)
{
- unsigned long *pir_ptr = gpci.posted_irq_desc;
- int i;
- fprintf(stderr, "-------Begin PIR dump-------\n");
- for (i = 0; i < 8; i++){
- fprintf(stderr, "Byte %d: 0x%016x\n", i, pir_ptr[i]);
+ Elf64_Ehdr *ehdr;
+ Elf *elf;
+ size_t phnum = 0;
+ Elf64_Phdr *hdrs;
+ int fd;
+
+ elf_version(EV_CURRENT);
+ fd = open(filename, O_RDONLY);
+ if (fd < 0) {
+ fprintf(stderr, "Can't open %s: %r\n", filename);
+ return 0;
+ }
+
+ elf = elf_begin(fd, ELF_C_READ, NULL);
+ if (elf == NULL) {
+ fprintf(stderr, "%s: cannot read %s ELF file.\n", __func__,
filename);
+ close(fd);
+ return 0;
+ }
+
+ ehdr = elf64_getehdr(elf);
+ if (ehdr == NULL) {
+ fprintf(stderr, "%s: cannot get exec header of %s.\n",
+ __func__, filename);
+ goto fail;
+ }
+ fprintf(stderr, "%s ELF entry point is %p\n", filename, ehdr->e_entry);
+
+ if (elf_getphdrnum(elf, &phnum) < 0) {
+ fprintf(stderr, "%s: cannot get program header num of %s.\n",
+ __func__, filename);
+ goto fail;
}
- fprintf(stderr, "-------End PIR dump-------\n");
+ fprintf(stderr, "%s has %d program headers\n", filename, phnum);
+
+ hdrs = elf64_getphdr(elf);
+ if (hdrs == NULL) {
+ fprintf(stderr, "%s: cannot get program headers of %s.\n",
+ __func__, filename);
+ goto fail;
+ }
+
+ for (int i = 0; i < phnum; i++) {
+ size_t tot;
+ Elf64_Phdr *h = &hdrs[i];
+ uintptr_t pa;
+
+ fprintf(stderr,
+ "%d: type 0x%lx flags 0x%lx offset 0x%lx vaddr 0x%lx
paddr 0x%lx size 0x%lx memsz 0x%lx align 0x%lx\n",
+ i,
+ h->p_type, /* Segment type */
+ h->p_flags, /* Segment flags */
+ h->p_offset, /* Segment file offset */
+ h->p_vaddr, /* Segment virtual address */
+ h->p_paddr, /* Segment physical address */
+ h->p_filesz, /* Segment size in file */
+ h->p_memsz, /* Segment size in memory */
+ h->p_align /* Segment alignment */);
+ if (h->p_type != PT_LOAD)
+ continue;
+ if ((h->p_flags & (PF_R|PF_W|PF_X)) == 0)
+ continue;
+
+ /* we do the memset purely to ensure everything gets paged in.
*/
+ /* compute the offset from the desired address. */
+ /* this ONLY works now if kernaddr > h->p_paddr */
+ pa = h->p_paddr;
+ memset((void *)pa, 0, h->p_memsz);
+ if (*kernstart > pa)
+ *kernstart = pa;
+ if (*kernend < pa+h->p_memsz)
+ *kernend = pa+h->p_memsz;
+ fprintf(stderr,
+ "Read header %d @offset %p to %p (elf PA is %p) %d
bytes:",
+ i, h->p_offset, pa, h->p_paddr, h->p_filesz);
+ tot = 0;
+ while (tot < h->p_filesz) {
+ int amt = pread(fd, (void *)(pa + tot), h->p_filesz -
tot,
+ h->p_offset + tot);
+ if (amt < 1)
+ break;
+ tot += amt;
+ }
+ fprintf(stderr, "read a total of %d bytes\n", tot);
+ if (tot < h->p_filesz) {
+ fprintf(stderr, "%s: got %d bytes, wanted %d bytes\n",
+ filename, tot, h->p_filesz);
+ goto fail;
+ }
+ }
+
+ close(fd);
+ elf_end(elf);
+ return ehdr->e_entry;
+ fail:
+ close(fd);
+ elf_end(elf);
+ return 0;
}
int main(int argc, char **argv)
@@ -347,11 +445,9 @@ int main(int argc, char **argv)
struct acpi_table_fadt *f;
struct acpi_table_madt *m;
struct acpi_table_xsdt *x;
- // lowmem is a bump allocated pointer to 2M at the "physbase" of memory
- void *lowmem = (void *) 0x1000000;
int amt;
int vmmflags = 0; // Disabled probably forever. VMM_VMCALL_PRINTF;
- uint64_t entry = 0x1200000, kerneladdress = 0x1200000;
+ uint64_t entry = 0;
int ret;
uintptr_t size;
void * xp;
@@ -370,6 +466,7 @@ int main(int argc, char **argv)
struct stat stat_result;
int num_read;
int option_index;
+ uintptr_t kernstart = (uintptr_t)~1, kernend = 0;
static struct option long_options[] = {
{"debug", no_argument, 0, 'd'},
{"vmm_vmcall", no_argument, 0, 'v'},
@@ -394,7 +491,6 @@ int main(int argc, char **argv)
exit(1);
}
memset(_kernel, 0, sizeof(_kernel));
- memset(lowmem, 0xff, 2*1048576);
vm->low4k = malloc(PGSIZE);
memset(vm->low4k, 0xff, PGSIZE);
vm->low4k[0x40e] = 0;
@@ -492,33 +588,16 @@ int main(int argc, char **argv)
fprintf(stderr, "Usage: %s vmimage [-n (no vmcall printf)]
[coreboot_tables [loadaddress [entrypoint]]]\n", argv[0]);
exit(1);
}
+
if (argc > 1)
coreboot_tables = (void *) strtoull(argv[1], 0, 0);
- if (argc > 2)
- kerneladdress = strtoull(argv[2], 0, 0);
- if (argc > 3)
- entry = strtoull(argv[3], 0, 0);
- kfd = open(argv[0], O_RDONLY);
- if (kfd < 0) {
- perror(argv[0]);
+
+ entry = load_kernel(argv[0], &kernstart, &kernend);
+ if (entry == 0) {
+ fprintf(stderr, "Unable to load kernel %s\n", argv[0]);
exit(1);
}
- // read in the kernel, one 2M page at a time.
- xp = (void *)kerneladdress;
- for(;;) {
- amt = read(kfd, xp, PML2_PTE_REACH);
- if (amt < 0) {
- perror("read");
- exit(1);
- }
- if (amt == 0) {
- break;
- }
- xp += amt;
- }
- size = ROUNDUP((uintptr_t)xp - kerneladdress, PML2_PTE_REACH);
- fprintf(stderr, "Read in %d bytes\n", size);
- close(kfd);
+
// The low 1m so we can fill in bullshit like ACPI. */
// And, sorry, due to the STUPID format of the RSDP for now we need the
low 1M.
@@ -530,7 +609,7 @@ int main(int argc, char **argv)
}
memset(low1m, 0xff, MiB-4096);
r = a;
- fprintf(stderr, "install rsdp to %p\n", r);
+ DBG("install rsdp to %p\n", r);
*r = rsdp;
a += sizeof(*r);
r->xsdt_physical_address = (uint64_t)a;
@@ -553,21 +632,21 @@ int main(int argc, char **argv)
x = a;
a += sizeof(*x) + 8*sizeof(void *);
memset(x, 0, a - (void *)x);
- fprintf(stderr, "install xsdt to %p\n", x);
+ DBG("install xsdt to %p\n", x);
*x = xsdt;
x->table_offset_entry[0] = 0;
x->table_offset_entry[1] = 0;
x->header.length = a - (void *)x;
f = a;
- fprintf(stderr, "install fadt to %p\n", f);
+ DBG("install fadt to %p\n", f);
*f = fadt;
x->table_offset_entry[0] = (uint64_t)f; // fadt MUST be first in xsdt!
a += sizeof(*f);
f->header.length = a - (void *)f;
f->Xdsdt = (uint64_t) a;
- fprintf(stderr, "install dsdt to %p\n", a);
+ DBG("install dsdt to %p\n", a);
memcpy(a, &DSDT_DSDTTBL_Header, 36);
a += 36;
@@ -581,7 +660,7 @@ int main(int argc, char **argv)
*m = madt;
x->table_offset_entry[3] = (uint64_t) m;
a += sizeof(*m);
- fprintf(stderr, "install madt to %p\n", m);
+ DBG("install madt to %p\n", m);
memmove(a, &Apic0, sizeof(Apic0));
a += sizeof(Apic0);
memmove(a, &Apic1, sizeof(Apic1));
@@ -606,7 +685,10 @@ int main(int argc, char **argv)
fprintf(stderr, "allchecksums ok\n");
- hexdump(stdout, r, a-(void *)r);
+ if (debug) {
+ fprintf(stderr, "ACPI tables:\n");
+ hexdump(stderr, r, a-(void *)r);
+ }
a = (void *)(((unsigned long)a + 0xfff) & ~0xfff);
gpci.posted_irq_desc = a;
@@ -722,7 +804,10 @@ int main(int argc, char **argv)
vm->nr_gpcs = 1;
vm->gpcis = &gpci;
ret = vmm_init(vm, vmmflags);
- assert(!ret);
+ if (ret) {
+ fprintf(stderr, "vmm_init failed, returns %d: %r\n", ret);
+ assert(!ret);
+ }
/* Allocate 3 pages for page table pages: a page of 512 GiB
* PTEs with only one entry filled to point to a page of 1 GiB
@@ -746,16 +831,21 @@ int main(int argc, char **argv)
p1 = &p512[NPTENTRIES];
p2m = &p512[2 * NPTENTRIES];
- p512[PML4(kerneladdress)] = (uint64_t)p1 | PTE_KERN_RW;
- p1[PML3(kerneladdress)] = (uint64_t)p2m | PTE_KERN_RW;
+ size = kernend - kernstart;
+ fprintf(stderr, "Map %p for %zu bytes\n", kernstart, size);
+ p512[PML4(kernstart)] = (uint64_t)p1 | PTE_KERN_RW;
+ p1[PML3(kernstart)] = (uint64_t)p2m | PTE_KERN_RW;
for (uintptr_t i = 0; i < size; i += PML2_PTE_REACH) {
- p2m[PML2(kerneladdress + i)] =
- (uint64_t)(kerneladdress + i) | PTE_KERN_RW | PTE_PS;
+ p2m[PML2(kernstart + i)] =
+ (uint64_t)(kernstart + i) | PTE_KERN_RW | PTE_PS;
}
uint8_t *kernel = (void *)GKERNBASE;
//write_coreboot_table(coreboot_tables, ((void *)VIRTIOBASE)
/*kernel*/, KERNSIZE + 1048576);
- hexdump(stdout, coreboot_tables, 512);
+ if (debug) {
+ fprintf(stderr, "Coreboot tables\n");
+ hexdump(stderr, coreboot_tables, 512);
+ }
fprintf(stderr, "p512 %p p512[0] is 0x%lx p1 %p p1[0] is 0x%x\n", p512,
p512[0], p1, p1[0]);
vmm_run_task(vm, timer_thread, 0);
--
2.8.0.rc3.226.g39d4020
--
You received this message because you are subscribed to the Google Groups
"Akaros" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
For more options, visit https://groups.google.com/d/optout.
