|author||Shao Miller <Shao.Miller@yrdsb.edu.on.ca>||2009-12-07 19:38:19 -0500|
|committer||H. Peter Anvin <email@example.com>||2009-12-07 19:56:59 -0800|
doc: document mBFT and "safe hook"syslinux-3.84-pre6
1 files changed, 40 insertions, 0 deletions
diff --git a/doc/memdisk.txt b/doc/memdisk.txt
index 8a308f10..254cb7c9 100644
@@ -9,6 +9,9 @@ disk and a (very small - 2K typical) chunk of low (DOS) memory for the
driver itself, then hooking the INT 13h (disk driver) and INT 15h
(memory query) BIOS interrupts.
+MEMDISK allows for an OS to detect the MEMDISK instance. (See the
+"Additional technical information" section below.)
To use it, type on the Syslinux command line:
@@ -241,3 +244,40 @@ method for this.
mov byte [es:bx], 0EAh ; FAR JMP
mov [es:bx+1], eax
+MEMDISK supports the Win9x "safe hook" structure for OS detection.
+(See "Safe Master Boot Record INT 13h Hook Routines," available at
+http://www.osronline.com/ddkx/w98ddk/storage_5l6g.htm as of
+December 7th, 2009.) An OS driver can take a look at the INTerrupt table
+and try to walk along the chain of those hooks that implement the "safe hook"
+structure. For each hook discovered, a vendor can be identified and the OS
+driver can take appropriate action. The OS driver can mark the "flags" field
+of the "safe hook" to indicate that the driver has reviewed it already. This
+prevents accidental re-detection, for example.
+MEMDISK adds one additional extension field to the "safe hook" structure, a
+pointer to a special MEMDISK structure called the "mBFT." The mBFT is the
+"MEMDISK Boot Firmware Table" (akin to the iSCSI iBFT and the AoE aBFT). An
+OS driver looking at MEMDISK's "safe hook" should know that this field will
+be present based on the fact that MEMDISK is the vendor identifier.
+The mBFT is little more than an ACPI table to prefix MEMDISK's traditional
+MEMDISK info structure (the "MDI"). The ACPI table's details are:
+ OEM ID. . . .: MEMDSK
+ OEM Table ID : Syslinux
+There is a 1-byte checksum field which covers the length of the mBFT all
+the way through to the end of the MEMDISK info structure.
+There is also a physical pointer to the "safe hook" structure associated
+with the MEMDISK instance. An OS driver might use the following logic:
+ 1. Walk INT 13h "safe hook" chain as far as possible, marking hooks as
+ having been reviewed. For MEMDISK hooks, the driver then follows the
+ pointer to the mBFT and gathers the RAM disk details from the included
+ 2. The OS driver scans low memory for valid mBFTs. MEMDISK instances that
+ have been "disconnected" from the INT 13h "safe hook" chain can be thus
+ discovered. Looking at their associated "safe hook" structure will
+ reveal if they were indeed reviewed by the previous stage.