Re: [PATCH] mm, kasan: don't poison boot memory

[George Kennedy <george.kennedy@oracle.com>]

On 2/25/2021 12:33 PM, George Kennedy wrote:
>
>
> On 2/25/2021 11:07 AM, Mike Rapoport wrote:
>> On Thu, Feb 25, 2021 at 10:22:44AM -0500, George Kennedy wrote:
>>>>>>> On 2/24/2021 5:37 AM, Mike Rapoport wrote:
>>> Applied just your latest patch, but same failure.
>>>
>>> I thought there was an earlier comment (which I can't find now) that 
>>> stated
>>> that memblock_reserve() wouldn't reserve the page, which is what's 
>>> needed
>>> here.
>> Actually, I think that memblock_reserve() should be just fine, but it 
>> seems
>> I'm missing something in address calculation each time.
>>
>> What would happen if you stuck
>>
>>     memblock_reserve(0xbe453000, PAGE_SIZE);
>>
>> say, at the beginning of find_ibft_region()?
>
> Good news Mike!
>
> The above hack in yesterday's last patch works - 10 successful 
> reboots. See: "BE453" below for the hack.
>
> I'll modify the patch to use "table_desc->address" instead, which is 
> the physical address of the table.
>
> diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
> index 7bdc023..c118dd5 100644
> --- a/arch/x86/kernel/acpi/boot.c
> +++ b/arch/x86/kernel/acpi/boot.c
> @@ -1551,6 +1551,7 @@ void __init acpi_boot_table_init(void)
>      if (acpi_disabled)
>          return;
>
> +#if 0
>      /*
>       * Initialize the ACPI boot-time table parser.
>       */
> @@ -1558,6 +1559,7 @@ void __init acpi_boot_table_init(void)
>          disable_acpi();
>          return;
>      }
> +#endif
>
>      acpi_table_parse(ACPI_SIG_BOOT, acpi_parse_sbf);
>
> diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
> index 740f3bdb..b045ab2 100644
> --- a/arch/x86/kernel/setup.c
> +++ b/arch/x86/kernel/setup.c
> @@ -571,16 +571,6 @@ void __init reserve_standard_io_resources(void)
>
>  }
>
> -static __init void reserve_ibft_region(void)
> -{
> -    unsigned long addr, size = 0;
> -
> -    addr = find_ibft_region(&size);
> -
> -    if (size)
> -        memblock_reserve(addr, size);
> -}
> -
>  static bool __init snb_gfx_workaround_needed(void)
>  {
>  #ifdef CONFIG_PCI
> @@ -1033,6 +1023,12 @@ void __init setup_arch(char **cmdline_p)
>       */
>      find_smp_config();
>
> +    /*
> +     * Initialize the ACPI boot-time table parser.
> +     */
> +    if (acpi_table_init())
> +        disable_acpi();
> +
>      reserve_ibft_region();
>
>      early_alloc_pgt_buf();
> diff --git a/drivers/firmware/iscsi_ibft_find.c 
> b/drivers/firmware/iscsi_ibft_find.c
> index 64bb945..95fc1a6 100644
> --- a/drivers/firmware/iscsi_ibft_find.c
> +++ b/drivers/firmware/iscsi_ibft_find.c
> @@ -47,7 +47,25 @@
>  #define VGA_MEM 0xA0000 /* VGA buffer */
>  #define VGA_SIZE 0x20000 /* 128kB */
>
> -static int __init find_ibft_in_mem(void)
> +static void __init *acpi_find_ibft_region(void)
> +{
> +    int i;
> +    struct acpi_table_header *table = NULL;
> +    acpi_status status;
> +
> +    if (acpi_disabled)
> +        return NULL;
> +
> +    for (i = 0; i < ARRAY_SIZE(ibft_signs) && !ibft_addr; i++) {
> +        status = acpi_get_table(ibft_signs[i].sign, 0, &table);
> +        if (ACPI_SUCCESS(status))
> +            return table;
> +    }
> +
> +    return NULL;
> +}
> +
> +static void __init *find_ibft_in_mem(void)
>  {
>      unsigned long pos;
>      unsigned int len = 0;
> @@ -70,35 +88,52 @@ static int __init find_ibft_in_mem(void)
>                  /* if the length of the table extends past 1M,
>                   * the table cannot be valid. */
>                  if (pos + len <= (IBFT_END-1)) {
> -                    ibft_addr = (struct acpi_table_ibft *)virt;
>                      pr_info("iBFT found at 0x%lx.\n", pos);
> -                    goto done;
> +                    return virt;
>                  }
>              }
>          }
>      }
> -done:
> -    return len;
> +
> +    return NULL;
>  }
> +
> +static void __init *find_ibft(void)
> +{
> +    /* iBFT 1.03 section 1.4.3.1 mandates that UEFI machines will
> +     * only use ACPI for this */
> +    if (!efi_enabled(EFI_BOOT))
> +        return find_ibft_in_mem();
> +    else
> +        return acpi_find_ibft_region();
> +}
> +
>  /*
>   * Routine used to find the iSCSI Boot Format Table. The logical
>   * kernel address is set in the ibft_addr global variable.
>   */
> -unsigned long __init find_ibft_region(unsigned long *sizep)
> +void __init reserve_ibft_region(void)
>  {
> -    ibft_addr = NULL;
> +    struct acpi_table_ibft *table;
> +    unsigned long size;
>
> -    /* iBFT 1.03 section 1.4.3.1 mandates that UEFI machines will
> -     * only use ACPI for this */
> +    table = find_ibft();
> +    if (!table)
> +        return;
>
> -    if (!efi_enabled(EFI_BOOT))
> -        find_ibft_in_mem();
> -
> -    if (ibft_addr) {
> -        *sizep = PAGE_ALIGN(ibft_addr->header.length);
> -        return (u64)virt_to_phys(ibft_addr);
> -    }
> +    size = PAGE_ALIGN(table->header.length);
> +#if 0
> +printk(KERN_ERR "XXX reserve_ibft_region: table=%llx, 
> virt_to_phys(table)=%llx, size=%lx\n",
> +    (u64)table, virt_to_phys(table), size);
> +    memblock_reserve(virt_to_phys(table), size);
> +#else
> +printk(KERN_ERR "XXX reserve_ibft_region: table=%llx, 
> 0x00000000BE453000, size=%lx\n",
> +    (u64)table, size);
> +    memblock_reserve(0x00000000BE453000, size);
> +#endif
>
> -    *sizep = 0;
> -    return 0;
> +    if (efi_enabled(EFI_BOOT))
> +        acpi_put_table(&table->header);
> +    else
> +        ibft_addr = table;
>  }
> diff --git a/include/linux/iscsi_ibft.h b/include/linux/iscsi_ibft.h
> index b7b45ca..da813c8 100644
> --- a/include/linux/iscsi_ibft.h
> +++ b/include/linux/iscsi_ibft.h
> @@ -26,13 +26,9 @@
>   * mapped address is set in the ibft_addr variable.
>   */
>  #ifdef CONFIG_ISCSI_IBFT_FIND
> -unsigned long find_ibft_region(unsigned long *sizep);
> +void reserve_ibft_region(void);
>  #else
> -static inline unsigned long find_ibft_region(unsigned long *sizep)
> -{
> -    *sizep = 0;
> -    return 0;
> -}
> +static inline void reserve_ibft_region(void) {}
>  #endif
>
>  #endif /* ISCSI_IBFT_H */

Mike,

To get rid of the 0x00000000BE453000 hardcoding, I added the following 
patch to your above patch to get the iBFT table "address" to use with 
memblock_reserve():

diff --git a/drivers/acpi/acpica/tbfind.c b/drivers/acpi/acpica/tbfind.c
index 56d81e4..4bc7bf3 100644
--- a/drivers/acpi/acpica/tbfind.c
+++ b/drivers/acpi/acpica/tbfind.c
@@ -120,3 +120,34 @@
      (void)acpi_ut_release_mutex(ACPI_MTX_TABLES);
      return_ACPI_STATUS(status);
  }
+
+acpi_physical_address
+acpi_tb_find_table_address(char *signature)
+{
+    acpi_physical_address address = 0;
+    struct acpi_table_desc *table_desc;
+    int i;
+
+    ACPI_FUNCTION_TRACE(tb_find_table_address);
+
+printk(KERN_ERR "XXX acpi_tb_find_table_address: signature=%s\n", 
signature);
+
+    (void)acpi_ut_acquire_mutex(ACPI_MTX_TABLES);
+    for (i = 0; i < acpi_gbl_root_table_list.current_table_count; ++i) {
+        if (memcmp(&(acpi_gbl_root_table_list.tables[i].signature),
+               signature, ACPI_NAMESEG_SIZE)) {
+
+            /* Not the requested table */
+
+            continue;
+        }
+
+        /* Table with matching signature has been found */
+        table_desc = &acpi_gbl_root_table_list.tables[i];
+        address = table_desc->address;
+    }
+
+    (void)acpi_ut_release_mutex(ACPI_MTX_TABLES);
+printk(KERN_ERR "XXX acpi_tb_find_table_address(EXIT): address=%llx\n", 
address);
+    return address;
+}
diff --git a/drivers/firmware/iscsi_ibft_find.c 
b/drivers/firmware/iscsi_ibft_find.c
index 95fc1a6..0de70b4 100644
--- a/drivers/firmware/iscsi_ibft_find.c
+++ b/drivers/firmware/iscsi_ibft_find.c
@@ -28,6 +28,8 @@

  #include <asm/mmzone.h>

+extern acpi_physical_address acpi_tb_find_table_address(char *signature);
+
  /*
   * Physical location of iSCSI Boot Format Table.
   */
@@ -116,24 +118,32 @@ void __init reserve_ibft_region(void)
  {
      struct acpi_table_ibft *table;
      unsigned long size;
+    acpi_physical_address address;

      table = find_ibft();
      if (!table)
          return;

      size = PAGE_ALIGN(table->header.length);
+    address = acpi_tb_find_table_address(table->header.signature);
  #if 0
  printk(KERN_ERR "XXX reserve_ibft_region: table=%llx, 
virt_to_phys(table)=%llx, size=%lx\n",
      (u64)table, virt_to_phys(table), size);
      memblock_reserve(virt_to_phys(table), size);
  #else
-printk(KERN_ERR "XXX reserve_ibft_region: table=%llx, 
0x00000000BE453000, size=%lx\n",
-    (u64)table, size);
-    memblock_reserve(0x00000000BE453000, size);
+printk(KERN_ERR "XXX reserve_ibft_region: table=%llx, address=%llx, 
size=%lx\n",
+    (u64)table, address, size);
+    if (address)
+        memblock_reserve(address, size);
+    else
+        printk(KERN_ERR "%s: Can't find table address\n", __func__);
  #endif

-    if (efi_enabled(EFI_BOOT))
+    if (efi_enabled(EFI_BOOT)) {
+printk(KERN_ERR "XXX reserve_ibft_region: calling 
acpi_put_table(%llx)\n", (u64)&table->header);
          acpi_put_table(&table->header);
-    else
+    } else {
          ibft_addr = table;
+printk(KERN_ERR "XXX reserve_ibft_region: ibft_addr=%llx\n", 
(u64)ibft_addr);
+    }
  }

Debug from the above:
[    0.050646] ACPI: Early table checksum verification disabled
[    0.051778] ACPI: RSDP 0x00000000BFBFA014 000024 (v02 BOCHS )
[    0.052922] ACPI: XSDT 0x00000000BFBF90E8 00004C (v01 BOCHS BXPCFACP 
00000001      01000013)
[    0.054623] ACPI: FACP 0x00000000BFBF5000 000074 (v01 BOCHS BXPCFACP 
00000001 BXPC 00000001)
[    0.056326] ACPI: DSDT 0x00000000BFBF6000 00238D (v01 BOCHS BXPCDSDT 
00000001 BXPC 00000001)
[    0.058016] ACPI: FACS 0x00000000BFBFD000 000040
[    0.058940] ACPI: APIC 0x00000000BFBF4000 000090 (v01 BOCHS BXPCAPIC 
00000001 BXPC 00000001)
[    0.060627] ACPI: HPET 0x00000000BFBF3000 000038 (v01 BOCHS BXPCHPET 
00000001 BXPC 00000001)
[    0.062304] ACPI: BGRT 0x00000000BE49B000 000038 (v01 INTEL EDK2     
00000002      01000013)
[    0.063987] ACPI: iBFT 0x00000000BE453000 000800 (v01 BOCHS BXPCFACP 
00000000      00000000)
[    0.065683] XXX acpi_tb_find_table_address: signature=iBFT
[    0.066754] XXX acpi_tb_find_table_address(EXIT): address=be453000
[    0.067959] XXX reserve_ibft_region: table=ffffffffff240000, 
address=be453000, size=1000
[    0.069534] XXX reserve_ibft_region: calling 
acpi_put_table(ffffffffff240000)

Not sure if it's the right thing to do, but added 
"acpi_tb_find_table_address()" to return the physical address of a table 
to use with memblock_reserve().

virt_to_phys(table) does not seem to return the physical address for the 
iBFT table (it would be nice if struct acpi_table_header also had a 
"address" element for the physical address of the table).

Ran 10 successful boots with the above without failure.

George
>
>
> Debug from the above:
>
> [    0.020293] last_pfn = 0xbfedc max_arch_pfn = 0x400000000
> [    0.050778] ACPI: Early table checksum verification disabled
> [    0.056475] ACPI: RSDP 0x00000000BFBFA014 000024 (v02 BOCHS )
> [    0.057628] ACPI: XSDT 0x00000000BFBF90E8 00004C (v01 BOCHS 
> BXPCFACP 00000001      01000013)
> [    0.059341] ACPI: FACP 0x00000000BFBF5000 000074 (v01 BOCHS 
> BXPCFACP 00000001 BXPC 00000001)
> [    0.061043] ACPI: DSDT 0x00000000BFBF6000 00238D (v01 BOCHS 
> BXPCDSDT 00000001 BXPC 00000001)
> [    0.062740] ACPI: FACS 0x00000000BFBFD000 000040
> [    0.063673] ACPI: APIC 0x00000000BFBF4000 000090 (v01 BOCHS 
> BXPCAPIC 00000001 BXPC 00000001)
> [    0.065369] ACPI: HPET 0x00000000BFBF3000 000038 (v01 BOCHS 
> BXPCHPET 00000001 BXPC 00000001)
> [    0.067061] ACPI: BGRT 0x00000000BE49B000 000038 (v01 INTEL 
> EDK2     00000002      01000013)
> [    0.068761] ACPI: iBFT 0x00000000BE453000 000800 (v01 BOCHS 
> BXPCFACP 00000000      00000000)
> [    0.070461] XXX reserve_ibft_region: table=ffffffffff240000, 
> 0x00000000BE453000, size=1000
> [    0.072231] check: Scanning 1 areas for low memory corruption
>
> George
>>> [   30.308229] iBFT detected..
>>> [   30.308796]
>>> ==================================================================
>>> [   30.308890] BUG: KASAN: use-after-free in ibft_init+0x134/0xc33
>>> [   30.308890] Read of size 4 at addr ffff8880be453004 by task 
>>> swapper/0/1
>>> [   30.308890]
>>> [   30.308890] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 
>>> 5.11.0-f9593a0 #12
>>> [   30.308890] Hardware name: QEMU Standard PC (i440FX + PIIX, 
>>> 1996), BIOS
>>> 0.0.0 02/06/2015
>>> [   30.308890] Call Trace:
>>> [   30.308890]  dump_stack+0xdb/0x120
>>> [   30.308890]  ? ibft_init+0x134/0xc33
>>> [   30.308890] print_address_description.constprop.7+0x41/0x60
>>> [   30.308890]  ? ibft_init+0x134/0xc33
>>> [   30.308890]  ? ibft_init+0x134/0xc33
>>> [   30.308890]  kasan_report.cold.10+0x78/0xd1
>>> [   30.308890]  ? ibft_init+0x134/0xc33
>>> [   30.308890]  __asan_report_load_n_noabort+0xf/0x20
>>> [   30.308890]  ibft_init+0x134/0xc33
>>> [   30.308890]  ? write_comp_data+0x2f/0x90
>>> [   30.308890]  ? ibft_check_initiator_for+0x159/0x159
>>> [   30.308890]  ? write_comp_data+0x2f/0x90
>>> [   30.308890]  ? ibft_check_initiator_for+0x159/0x159
>>> [   30.308890]  do_one_initcall+0xc4/0x3e0
>>> [   30.308890]  ? perf_trace_initcall_level+0x3e0/0x3e0
>>> [   30.308890]  ? unpoison_range+0x14/0x40
>>> [   30.308890]  ? ____kasan_kmalloc.constprop.5+0x8f/0xc0
>>> [   30.308890]  ? kernel_init_freeable+0x420/0x652
>>> [   30.308890]  ? __kasan_kmalloc+0x9/0x10
>>> [   30.308890]  ? __sanitizer_cov_trace_pc+0x21/0x50
>>> [   30.308890]  kernel_init_freeable+0x596/0x652
>>> [   30.308890]  ? console_on_rootfs+0x7d/0x7d
>>> [   30.308890]  ? __sanitizer_cov_trace_pc+0x21/0x50
>>> [   30.308890]  ? rest_init+0xf0/0xf0
>>> [   30.308890]  kernel_init+0x16/0x1d0
>>> [   30.308890]  ? rest_init+0xf0/0xf0
>>> [   30.308890]  ret_from_fork+0x22/0x30
>>> [   30.308890]
>>> [   30.308890] The buggy address belongs to the page:
>>> [   30.308890] page:0000000001b7b17c refcount:0 mapcount:0
>>> mapping:0000000000000000 index:0x1 pfn:0xbe453
>>> [   30.308890] flags: 0xfffffc0000000()
>>> [   30.308890] raw: 000fffffc0000000 ffffea0002ef9788 ffffea0002f91488
>>> 0000000000000000
>>> [   30.308890] raw: 0000000000000001 0000000000000000 00000000ffffffff
>>> 0000000000000000
>>> [   30.308890] page dumped because: kasan: bad access detected
>>> [   30.308890] page_owner tracks the page as freed
>>> [   30.308890] page last allocated via order 0, migratetype Movable,
>>> gfp_mask 0x100dca(GFP_HIGHUSER_MOVABLE|__GFP_ZERO), pid 204, ts 
>>> 28121288605
>>> [   30.308890]  prep_new_page+0xfb/0x140
>>> [   30.308890]  get_page_from_freelist+0x3503/0x5730
>>> [   30.308890]  __alloc_pages_nodemask+0x2d8/0x650
>>> [   30.308890]  alloc_pages_vma+0xe2/0x560
>>> [   30.308890]  __handle_mm_fault+0x930/0x26c0
>>> [   30.308890]  handle_mm_fault+0x1f9/0x810
>>> [   30.308890]  do_user_addr_fault+0x6f7/0xca0
>>> [   30.308890]  exc_page_fault+0xaf/0x1a0
>>> [   30.308890]  asm_exc_page_fault+0x1e/0x30
>>> [   30.308890] page last free stack trace:
>>> [   30.308890]  free_pcp_prepare+0x122/0x290
>>> [   30.308890]  free_unref_page_list+0xe6/0x490
>>> [   30.308890]  release_pages+0x2ed/0x1270
>>> [   30.308890]  free_pages_and_swap_cache+0x245/0x2e0
>>> [   30.308890]  tlb_flush_mmu+0x11e/0x680
>>> [   30.308890]  tlb_finish_mmu+0xa6/0x3e0
>>> [   30.308890]  exit_mmap+0x2b3/0x540
>>> [   30.308890]  mmput+0x11d/0x450
>>> [   30.308890]  do_exit+0xaa6/0x2d40
>>> [   30.308890]  do_group_exit+0x128/0x340
>>> [   30.308890]  __x64_sys_exit_group+0x43/0x50
>>> [   30.308890]  do_syscall_64+0x37/0x50
>>> [   30.308890]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
>>> [   30.308890]
>>> [   30.308890] Memory state around the buggy address:
>>> [   30.308890]  ffff8880be452f00: ff ff ff ff ff ff ff ff ff ff ff 
>>> ff ff ff
>>> ff ff
>>> [   30.308890]  ffff8880be452f80: ff ff ff ff ff ff ff ff ff ff ff 
>>> ff ff ff
>>> ff ff
>>> [   30.308890] >ffff8880be453000: ff ff ff ff ff ff ff ff ff ff ff 
>>> ff ff ff
>>> ff ff
>>> [   30.308890]                    ^
>>> [   30.308890]  ffff8880be453080: ff ff ff ff ff ff ff ff ff ff ff 
>>> ff ff ff
>>> ff ff
>>> [   30.308890]  ffff8880be453100: ff ff ff ff ff ff ff ff ff ff ff 
>>> ff ff ff
>>> ff ff
>>> [   30.308890]
>>> ==================================================================
>>>
>>> George
>>>
>