25 Коммитов

Автор	SHA1	Сообщение	Дата
Andrew Baumann	1238e67a9e	global rename from kevlar to komodo	2016-08-30 10:43:21 -07:00
Andrew Baumann	3937cb9bbe	pass secure_physbase to monitor by setting it directly from the loader ... This avoids a special-case SMC call	2016-06-20 12:56:55 -07:00
Andrew Baumann	58dad79479	setup mvbar distinct from secure vbar	2016-06-14 15:52:39 -07:00
Andrew Baumann	a7e674f472	further simplify the monitor <-> bootloader interface, so the initial setup call is just another SMC	2016-06-07 16:31:11 -07:00
Andrew Baumann	a562290644	simplify monitor startup, by moving code to leave secure world into ... the loader and setting up sp_mon from there	2016-06-06 16:14:47 -07:00
Andrew Baumann	7b83f963f1	initial implementation of address space switch	2016-06-02 13:15:57 -07:00
Andrew Baumann	cf6cc7e98b	clean up monitor's virtual memory layout, so that everything is statically allocated ... change user-mode VM region to 1GB, so all allocations can be 4kB (including the L1 tables) added direct-map region, so we can look at by-reference parameters from normal world	2016-05-24 15:34:14 -07:00
Andrew Baumann	d9e0fc37df	checkpoint: using loaderblock in monitor, sizing the page DB, handling trivial SMCs	2016-05-13 16:39:20 -07:00
Andrew Baumann	d02b208626	don't forget to stack LR across SMC C handler; now back into Linux boot	2016-05-13 09:49:45 -07:00
Andrew Baumann	5b04b1c335	checkpoint: started work to pass a loader block to the monitor, and have it always responsible for leaving secure world ... currently broken, because of a problem with qemu's MSR/MRS instructions that I need to track down	2016-05-12 16:53:53 -07:00
Andrew Baumann	947fb44544	fix alignment bug (and dumb loop) when mapping monitor with a non-empty BSS	2016-05-12 16:32:16 -07:00
Andrew Baumann	7ebda9435e	tweak smc path to avoid using/specifying the immediate, per the ARM SMC ABI spec	2016-04-29 12:14:58 -07:00
Andrew Baumann	5017bee158	SMC test code ... * avoid trampling r4 * changed from -march=armv7-a to -mcpu=cortex-a7 to enable SMC instruction (why?)	2016-04-19 16:42:46 +01:00
Andrew Baumann	a4eeeec048	enable caches (!) and sharing in PTEs	2016-04-19 11:53:57 +01:00
Andrew Baumann	3a3a8bc386	debugging goop to try to debug hw issue (hang when enabling MMU); no luck yet	2016-04-12 16:25:44 -07:00
Andrew Baumann	a89c1db659	checkpoint: entering linux, and parking other cores	2016-04-10 20:29:41 -07:00
Andrew Baumann	b960572817	now calling into and returning to loader from monitor (in SW PL1)	2016-04-08 21:29:05 -07:00
Andrew Baumann	963cb607f0	checkpoint: fixed some bugs, into early monitor binary (bad stack addr?) with translations on	2016-04-08 16:27:58 -07:00
Andrew Baumann	28a6e1c584	checkpoint: as far as calling into the monitor on qemu, but translations don't seem to work	2016-04-06 16:52:33 -07:00
Andrew Baumann	1e569e9e97	checkpoint: lots of work on mapping monitor, but none of it tested	2016-04-05 16:36:56 -07:00
Andrew Baumann	fe562b0633	add code to reserve high phys mem for the monitor and adjust size reported to linux	2016-04-04 16:49:46 -07:00
Andrew Baumann	f328cf3df3	checkpoint: booting on hw again, looking at atags	2016-04-04 16:18:53 -07:00
Andrew Baumann	95574c16fb	build/link three different blobs (loader, kernel, monitor)	2016-04-01 16:56:13 -07:00
Andrew Baumann	7c25a59e17	setup all cores, and prep for loading a separate secure monitor	2016-04-01 14:53:26 -07:00
Andrew Baumann	a96ba7e116	refactor into subdirectories, after thinking through the components needed ... cleaned up build system a tad	2016-03-29 16:16:32 -07:00