2,952
edits
Changes
Jump to navigation
Jump to search
Line 287:
Line 287: − Line 314:
Line 313: + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Line 1,530:
Line 1,573: − * MC_SECURITY_CFG0+ − * MC_SECURITY_CFG1+ − * MC_SECURITY_CFG3+ + + Line 1,538:
Line 1,583: − * MC_VIDEO_PROTECT_GPU_OVERRIDE_0+ − * MC_VIDEO_PROTECT_GPU_OVERRIDE_1+ − * MC_VIDEO_PROTECT_BOM+ − * MC_VIDEO_PROTECT_SIZE_MB+ − * MC_VIDEO_PROTECT_REG_CTRL+ + + Line 1,548:
Line 1,595: − * MC_SEC_CARVEOUT_BOM+ − * MC_SEC_CARVEOUT_SIZE_MB+ − * MC_SEC_CARVEOUT_REG_CTRL+ + + Line 1,556:
Line 1,605: − * MC_MTS_CARVEOUT_BOM+ − * MC_MTS_CARVEOUT_SIZE_MB+ − * MC_MTS_CARVEOUT_ADR_HI+ − * MC_MTS_CARVEOUT_REG_CTRL+ + + − + − + − + − + − + − + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + − However, in the Switch's case they are currently assigned as:+ − + − + − + − + − + + + + + + + + + + + + − These carveouts are controlled by the following MC registers:+ − + − + − + − + − + − + − + − + − + − + − + − + − + − + + + + + + + + + − + − + − Granule size for TTBR0*_EL1 is 4KB.+ − TTBR0_EL1 vmem starts at vaddr 0x0.+ − vmem end-addr for TTBR1_EL1 is 0xffffffffffffffff. vmem start-addr for TTBR1_EL1 is 0xFFFFFFF000000000.+ − T0SZ = 31. Hence, bit-size of the TTBR0*_EL1 vmem region is 33. (0x0000000200000000)+ − T1SZ = 28. Hence, bit-size of the TTBR1*_EL1 vmem region is 36. (0x0000001000000000)+ − + − Note: ARM config for TTBR0 is presumably configured for userland later.+ − + − See arm-doc for "Table D4-25 Translation table entry addresses when using the 4KB translation granule".+ − + − See arm-doc for "Overview of VMSAv8-64 address translation using the 4KB translation granule".+ − + − See arm-doc for "Table D4-11 TCR.TnSZ values and IA ranges, 4K granule with no concatenation of tables".+ − Both TTBR*_EL1 use "Initial lookup level" 1. Therefore, the TTBR*_EL1 tables are level1.+ − + − Due to T*SZ, Stage1/Stage2 translation for the initial table(level1) are the same, except Stage2 uses hard-coded T0SZ. − Basically, the table is accessed as: ((u64*)tablebase)[<IA[y:30]>], where y = (37-T*SZ)+26. That is, starting at bit "y" ending(inclusive) at bit30. For TTBR0*_EL1, y = 32, while for TTBR1_EL1 y = 35. − Hence, for TTBR0, index=((vaddr>>30) & 0x7), and for TTBR1, index=((vaddr>>30) & 0x3f). − "Vector Base Address Register (EL1)" = 0xfffffff7ffc50800.+ − The table for TTBR0 only contains the following:+ − * Vmem 0x80000000 is mapped to physmem 0x80000000, using a size loaded from a register. This is only done when: "endaddr = 0x7fffffff + size; if(endaddr >= 0x80000001){...}"+ − ** The size is loaded from: "(u32 *0x70019050 & 0x3fff) << 20;" − ** The value written to the MMU-table descriptor is: "physaddr | val | 0x709;". val is 1<<52 when "tmp>>34" is non-zero and when "if((physaddr & 0x3c0000000) == 0)", otherwise val=0. tmp=size at the start and increased by 0xffffffffc0000000 each loop iteration. physaddr is increased by 0x40000000 each loop iteration. − TTBR1:+ − + − + + + + + + + + + + + + + + + − * Initializes level2 pagetable descriptor for vmem 0xFFFFFFF7C0000000. descriptor = 0x3 | physaddr. physaddr is core-specific.+ − * Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FFC00000. descriptor = 0x3 | physaddr. physaddr is core-specific. − * The content of the pagetable for the following level3 mmutables are not initialized in the main mmutable-init func. descriptor = 0x8007c003(0x3 | <physaddr tablebase>). tablebase=0x8007c000. − ** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FEE00000. physaddr = tablebase + (0x1<<12). − ** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF000000. physaddr = tablebase + (0x2<<12). − ** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF200000. physaddr = tablebase + (0x3<<12). − ** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FFA00000. physaddr = tablebase + (0x7<<12). − ** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FEC00000. physaddr = tablebase. − ** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF400000. physaddr = tablebase + (0x4<<12). − ** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF600000. physaddr = tablebase + (0x5<<12). − ** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF800000. physaddr = tablebase + (0x6<<12).
no edit summary
| All || 0xFFFFFFF7FFDFD000-0xFFFFFFF7FFDFDFFF || 0x50042000 || 0x1000 || 0x60000000000607 || RW- || Interrupt Controller Physical CPU interface
| All || 0xFFFFFFF7FFDFD000-0xFFFFFFF7FFDFDFFF || 0x50042000 || 0x1000 || 0x60000000000607 || RW- || Interrupt Controller Physical CPU interface
|}
|}
The rest are are mapped to core-specific physaddrs, each one is 0x1000-bytes. Descriptor ORR-value = 0x6000000000070B.
The rest are are mapped to core-specific physaddrs, each one is 0x1000-bytes. Descriptor ORR-value = 0x6000000000070B.
| 0x800XX000
| 0x800XX000
|}
|}
== Notes ==
=== 2.0.0 ===
Granule size for TTBR0*_EL1 is 4KB.
TTBR0_EL1 vmem starts at vaddr 0x0.
vmem end-addr for TTBR1_EL1 is 0xffffffffffffffff. vmem start-addr for TTBR1_EL1 is 0xFFFFFFF000000000.
T0SZ = 31. Hence, bit-size of the TTBR0*_EL1 vmem region is 33. (0x0000000200000000)
T1SZ = 28. Hence, bit-size of the TTBR1*_EL1 vmem region is 36. (0x0000001000000000)
Note: ARM config for TTBR0 is presumably configured for userland later.
See arm-doc for "Table D4-25 Translation table entry addresses when using the 4KB translation granule".
See arm-doc for "Overview of VMSAv8-64 address translation using the 4KB translation granule".
See arm-doc for "Table D4-11 TCR.TnSZ values and IA ranges, 4K granule with no concatenation of tables".
Both TTBR*_EL1 use "Initial lookup level" 1. Therefore, the TTBR*_EL1 tables are level1.
Due to T*SZ, Stage1/Stage2 translation for the initial table(level1) are the same, except Stage2 uses hard-coded T0SZ.
Basically, the table is accessed as: ((u64*)tablebase)[<IA[y:30]>], where y = (37-T*SZ)+26. That is, starting at bit "y" ending(inclusive) at bit30. For TTBR0*_EL1, y = 32, while for TTBR1_EL1 y = 35.
Hence, for TTBR0, index=((vaddr>>30) & 0x7), and for TTBR1, index=((vaddr>>30) & 0x3f).
"Vector Base Address Register (EL1)" = 0xfffffff7ffc50800.
The table for TTBR0 only contains the following:
* Vmem 0x80000000 is mapped to physmem 0x80000000, using a size loaded from a register. This is only done when: "endaddr = 0x7fffffff + size; if(endaddr >= 0x80000001){...}"
** The size is loaded from: "(u32 *0x70019050 & 0x3fff) << 20;"
** The value written to the MMU-table descriptor is: "physaddr | val | 0x709;". val is 1<<52 when "tmp>>34" is non-zero and when "if((physaddr & 0x3c0000000) == 0)", otherwise val=0. tmp=size at the start and increased by 0xffffffffc0000000 each loop iteration. physaddr is increased by 0x40000000 each loop iteration.
TTBR1:
* vmem 0xFFFFFFF800000000 is mapped to physmem 0x80000000. Similar to above, except tmp=0 due to wrap-around, etc. This also has usermode/kernel XN enabled in the descriptor ORR-value. The chunksize used when increasing addr is 0xfffffff840000000, with another +=0x40000000 separate from the addr cmp for the loop.
** "endaddr = 0x3fffffff + (<size from above> | 0xfffffff800000000); enaddr = (endaddr & 0xffffffffc0000000)-1; if(endaddr >= 0xfffffff800000001){<map mem>}"
* Initializes level2 pagetable descriptor for vmem 0xFFFFFFF7C0000000. descriptor = 0x3 | physaddr. physaddr is core-specific.
* Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FFC00000. descriptor = 0x3 | physaddr. physaddr is core-specific.
* The content of the pagetable for the following level3 mmutables are not initialized in the main mmutable-init func. descriptor = 0x8007c003(0x3 | <physaddr tablebase>). tablebase=0x8007c000.
** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FEE00000. physaddr = tablebase + (0x1<<12).
** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF000000. physaddr = tablebase + (0x2<<12).
** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF200000. physaddr = tablebase + (0x3<<12).
** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FFA00000. physaddr = tablebase + (0x7<<12).
** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FEC00000. physaddr = tablebase.
** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF400000. physaddr = tablebase + (0x4<<12).
** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF600000. physaddr = tablebase + (0x5<<12).
** Initializes level3 pagetable descriptor for vmem 0xFFFFFFF7FF800000. physaddr = tablebase + (0x6<<12).
= Secure Monitor =
= Secure Monitor =
This carveout is controlled by the following MC registers:
This carveout is controlled by the following MC registers:
<pre>
MC_SECURITY_CFG0
MC_SECURITY_CFG1
MC_SECURITY_CFG3
</pre>
== VPR Carveout ==
== VPR Carveout ==
This carveout is controlled by the following MC registers:
This carveout is controlled by the following MC registers:
<pre>
MC_VIDEO_PROTECT_GPU_OVERRIDE_0
MC_VIDEO_PROTECT_GPU_OVERRIDE_1
MC_VIDEO_PROTECT_BOM
MC_VIDEO_PROTECT_SIZE_MB
MC_VIDEO_PROTECT_REG_CTRL
</pre>
== SEC Carveout ==
== SEC Carveout ==
This carveout is controlled by the following MC registers:
This carveout is controlled by the following MC registers:
<pre>
MC_SEC_CARVEOUT_BOM
MC_SEC_CARVEOUT_SIZE_MB
MC_SEC_CARVEOUT_REG_CTRL
</pre>
== MTS Carveout ==
== MTS Carveout ==
This carveout is controlled by the following MC registers:
This carveout is controlled by the following MC registers:
<pre>
MC_MTS_CARVEOUT_BOM
MC_MTS_CARVEOUT_SIZE_MB
MC_MTS_CARVEOUT_ADR_HI
MC_MTS_CARVEOUT_REG_CTRL
</pre>
== Generalized Carveouts ==
== Generalized Carveouts ==
These carveouts can be freely configured for any client that supports them. By default, they are assigned as:
These carveouts can be freely configured for any client that supports them.
* GSC 1: NVDEC
* GSC 2: WPR1 (for GPU)
These carveouts are controlled by the following MC registers:
* GSC 3: WPR2 (for GPU)
<pre>
* GSC 4: TSECA
MC_SECURITY_CARVEOUT1/2/3/4/5_BOM
* GSC 5: TSECB
MC_SECURITY_CARVEOUT1/2/3/4/5_BOM_HI
MC_SECURITY_CARVEOUT1/2/3/4/5_SIZE_128KB
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS0
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS1
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS2
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS3
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS4
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS0
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS1
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS2
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS3
MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS4
MC_SECURITY_CARVEOUT1/2/3/4/5_CFG0
</pre>
=== GSC1 ===
This carveout is, by default, for NVDEC. In the Switch's case, this carveout is not used.
It is configured as follows:
<pre>
*(u32 *)MC_SECURITY_CARVEOUT1_BOM = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_BOM_HI = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_SIZE_128KB = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_ACCESS0 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_ACCESS1 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_ACCESS2 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_ACCESS3 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_ACCESS4 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_FORCE_INTERNAL_ACCESS0 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_FORCE_INTERNAL_ACCESS1 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_FORCE_INTERNAL_ACCESS2 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_FORCE_INTERNAL_ACCESS3 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CLIENT_FORCE_INTERNAL_ACCESS4 = 0;
*(u32 *)MC_SECURITY_CARVEOUT1_CFG0 = 0x4000006;
</pre>
=== GSC2 ===
This carveout is, by default and in the Switch's case, for the GPU (WPR1).
It is configured as follows:
* GSC 1: None
<pre>
* GSC 2: GPU
*(u32 *)MC_SECURITY_CARVEOUT2_BOM = 0x80020000;
* GSC 3: None
*(u32 *)MC_SECURITY_CARVEOUT2_BOM_HI = 0;
* GSC 4: Kernel (active)
*(u32 *)MC_SECURITY_CARVEOUT2_SIZE_128KB = 2;
* GSC 5: Kernel (unused)
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_ACCESS0 = 0;
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_ACCESS1 = 0;
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_ACCESS2 = 0x3000000;
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_ACCESS3 = 0;
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_ACCESS4 = 0x300;
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_FORCE_INTERNAL_ACCESS0 = 0;
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_FORCE_INTERNAL_ACCESS1 = 0;
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_FORCE_INTERNAL_ACCESS2 = 0;
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_FORCE_INTERNAL_ACCESS3 = 0;
*(u32 *)MC_SECURITY_CARVEOUT2_CLIENT_FORCE_INTERNAL_ACCESS4 = 0;
*(u32 *)MC_SECURITY_CARVEOUT2_CFG0 = 0x440167E;
</pre>
=== GSC3 ===
* MC_SECURITY_CARVEOUT1/2/3/4/5_BOM
This carveout is, by default, for the GPU (WPR2). In the Switch's case, this carveout is not used.
* MC_SECURITY_CARVEOUT1/2/3/4/5_BOM_HI
* MC_SECURITY_CARVEOUT1/2/3/4/5_SIZE_128KB
It is configured as follows:
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS0
<pre>
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS1
*(u32 *)MC_SECURITY_CARVEOUT3_BOM = 0;
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS2
*(u32 *)MC_SECURITY_CARVEOUT3_BOM_HI = 0;
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS3
*(u32 *)MC_SECURITY_CARVEOUT3_SIZE_128KB = 0;
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_ACCESS4
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_ACCESS0 = 0;
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS0
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_ACCESS1 = 0;
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS1
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_ACCESS2 = 0x3000000;
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS2
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_ACCESS3 = 0;
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS3
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_ACCESS4 = 0x300;
* MC_SECURITY_CARVEOUT1/2/3/4/5_CLIENT_FORCE_INTERNAL_ACCESS4
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_FORCE_INTERNAL_ACCESS0 = 0;
* MC_SECURITY_CARVEOUT1/2/3/4/5_CFG0
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_FORCE_INTERNAL_ACCESS1 = 0;
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_FORCE_INTERNAL_ACCESS2 = 0;
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_FORCE_INTERNAL_ACCESS3 = 0;
*(u32 *)MC_SECURITY_CARVEOUT3_CLIENT_FORCE_INTERNAL_ACCESS4 = 0;
*(u32 *)MC_SECURITY_CARVEOUT3_CFG0 = 0x4401E7E;
</pre>
=== GSC4 ===
This carveout is, by default, for TSECA. In the Switch's case, this carveout is used by the Kernel.
= Notes =
It is initially configured as follows:
== 2.0.0 ==
<pre>
*(u32 *)MC_SECURITY_CARVEOUT4_BOM = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_BOM_HI = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_SIZE_128KB = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_ACCESS0 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_ACCESS1 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_ACCESS2 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_ACCESS3 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_ACCESS4 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_FORCE_INTERNAL_ACCESS0 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_FORCE_INTERNAL_ACCESS1 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_FORCE_INTERNAL_ACCESS2 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_FORCE_INTERNAL_ACCESS3 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CLIENT_FORCE_INTERNAL_ACCESS4 = 0;
*(u32 *)MC_SECURITY_CARVEOUT4_CFG0 = 0x8F;
</pre>
Then further configured using [[SMC#ConfigureCarveout|smcConfigureCarveout]].
=== GSC5 ===
This carveout is, by default, for TSECB. In the Switch's case, this carveout is reserved for the Kernel.
It is initially configured as follows:
* vmem 0xFFFFFFF800000000 is mapped to physmem 0x80000000. Similar to above, except tmp=0 due to wrap-around, etc. This also has usermode/kernel XN enabled in the descriptor ORR-value. The chunksize used when increasing addr is 0xfffffff840000000, with another +=0x40000000 separate from the addr cmp for the loop.
<pre>
** "endaddr = 0x3fffffff + (<size from above> | 0xfffffff800000000); enaddr = (endaddr & 0xffffffffc0000000)-1; if(endaddr >= 0xfffffff800000001){<map mem>}"
*(u32 *)MC_SECURITY_CARVEOUT5_BOM = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_BOM_HI = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_SIZE_128KB = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_ACCESS0 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_ACCESS1 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_ACCESS2 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_ACCESS3 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_ACCESS4 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_FORCE_INTERNAL_ACCESS0 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_FORCE_INTERNAL_ACCESS1 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_FORCE_INTERNAL_ACCESS2 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_FORCE_INTERNAL_ACCESS3 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CLIENT_FORCE_INTERNAL_ACCESS4 = 0;
*(u32 *)MC_SECURITY_CARVEOUT5_CFG0 = 0x8F;
</pre>
It can be further configured using [[SMC#ConfigureCarveout|smcConfigureCarveout]], but it's currently unused.