Merge pull request #11 from CTSRD-CHERI/initial-fixes

Minor fixes and formatting changes

src/cheri-pte-ext.adoc

@@ -43,8 +43,8 @@ clears the tag bit of the capability written to a virtual page with the CW bit
 clear.
 
 NOTE: The implementation of the CW bit does not force a dependency on the tag
-bit's value of the capability written, so implementations must support this
-feature.
+bit's value of the capability written, so implementations must support the CW
+bit.
 
 The CD bit indicates that a capability with tag set has been written to the
 virtual page since the last time the CD bit was cleared. Implementations are
@@ -55,7 +55,7 @@ to manage the CD bit are permitted:
 instruction is executed, the <<pcc>> grants store capability permission, the
 tag bit of the capability being written is set and the address written
 corresponds to a virtual page with the CD bit clear.
-* When a capability store or AMO instruction is execute, the <<pcc>> grants store
+* When a capability store or AMO instruction is executed, the <<pcc>> grants store
 capability permission, the tag bit of the capability being written is set and
 the store address corresponds to a virtual page with the CD bit clear, the
 implementation sets the corresponding bit in the PTE. The PTE update must be

src/insns/cmove_cmv_16bit.adoc

@@ -12,22 +12,22 @@ Synopsis::
 Capability move (C.MV, C.CMOVE), 16-bit encoding
 
 Capability Mode Mnemonic::
-c.cmove cd, cs2`
+c.cmove cd, cs2
 
 Capability Mode Expansion::
-cmove cd, cs2`
+cmove cd, cs2
 
 Legacy Mode Mnemonic::
-c.mv rd, rs2`
+c.mv rd, rs2
 
 Legacy Mode Expansion::
-add rd, x0, rs2`
+add rd, x0, rs2
 
 Encoding::
 include::wavedrom/c_mv.adoc[]
 
 Capability Mode Description::
-Capability register cd is replaced with the contents of cs1.
+Capability register `cd` is replaced with the contents of `cs2`.
 
 Legacy Mode Description::
 Standard RISC-V C.MV instruction.

src/insns/zcmp_cmpopretz.adoc

@@ -9,7 +9,7 @@ See <<CM.CPOPRETZ>> and cite:[riscv-code-size-spec].
 ==== CM.CPOPRETZ
 
 Synopsis::
-Destroy stack frame (CM.CPOPRETZ, CM.POPRETZ): load the return address register and 0 to 12 saved registers from the stack frame, deallocate the stack frame. Move zero into argument register zero. Return through the return address register. 16-bit encodings.
+Destroy stack frame (CM.CPOPRETZ, CM.POPRETZ): load the return address register and register 0 to 12 saved registers from the stack frame, deallocate the stack frame. Move zero into argument register zero. Return through the return address register. 16-bit encodings.
 
 Capability Mode Mnemonic::
 `cm.cpopretz \{creg_list\}, -stack_adj`

src/instructions.adoc

@@ -17,7 +17,7 @@ jumps and conditional branches is bounds checked against <<pcc>> regardless of
 CHERI execution mode
 
 NOTE: Not all RISC-V extensions have been checked against CHERI. Compatible
-extensions, will eventually be listed in a CHERI profile.
+extensions will eventually be listed in a CHERI profile.
 
 <<<
 === "Zcheri_purecap", "Zcheri_legacy" and "Zcheri_mode" Extensions for CHERI

src/riscv-integration.adoc

@@ -32,11 +32,12 @@ where address 2^XLENMAX^ - 1 is within the bounds.
 === Programmer's Model for Zcheri_purecap
 
 For {cheri_base_ext_name}, the 32 unprivileged *x* registers of the base
-integer ISA are extended so that they are able to hold a capability. Therefore,
-each *x* register is CLEN bits wide and has an out of band tag bit. The *x*
-notation refers to the address field of the capability in an unprivileged
-register while the *c* notation is used to refer to the full capability (i.e.
-address, metadata and tag) held in the same unprivileged register.
+integer ISA are extended so that they are able to hold a capability as well
+as renamed to *c* registers. Therefore, each *c* register is CLEN bits wide
+and has an out of band tag bit. The *x* notation refers to the address field
+of the capability in an unprivileged register while the *c* notation is used
+to refer to the full capability (i.e. address, metadata and tag) held in the
+same unprivileged register.
 
 Register *c0* is hardwired with all bits, including the capability metadata and
 tag, equal to 0. In other words, *c0* is hardwired to the <<null-cap>>
@@ -117,7 +118,7 @@ output tag is always 0
 * <<CANDPERM>>: bitwise AND of a mask value with a bit map representation of the
 architectural (AP) and software-defined (SDP) permissions fields
 * <<CSETBOUNDS>>: set the base and length of a capability. The tag is
-cleared, if the encoding cannot represents the bounds exactly
+cleared, if the encoding cannot represent the bounds exactly
 * <<CSETBOUNDSINEXACT>>: set the base and length of a capability. The base will be
 rounded down and/or the length will be rounded up if the encoding cannot represent
 the bounds exactly
@@ -177,7 +178,7 @@ The indirect jump and link <<pcc>> (<<JALR_PCC>>) instruction allows uncondition
 jumps to a target address. The target address is provided in an *x* register;
 the new address is installed in the address field of the <<pcc>>. The address of
 the instruction following the jump (*pc* + 4) is written to an *x* register.
-<<JALR_PCC>> causes an exceptions when a minimum sized instruction at the
+<<JALR_PCC>> causes an exception when a minimum sized instruction at the
 target address is not within the bounds of the <<pcc>> or the target address is
 misaligned.
 
@@ -1057,7 +1058,7 @@ NOTE: `auth_cap` is <<ddc>> for Legacy mode and `cs1` for Capability Mode
 | all stores, all atomics, all cbos | {cheri_excep_mcause} | {cheri_excep_type_data} | {cheri_excep_cause_seal}   |`auth_cap` seal       | isCapSealed(`auth_cap`)
 |             all atomics, all cbos | {cheri_excep_mcause} | {cheri_excep_type_data} | {cheri_excep_cause_perm}   |`auth_cap` permission | AMO only: not(`auth_cap`.<<r_perm>>)
 | all stores, all atomics, all cbos | {cheri_excep_mcause} | {cheri_excep_type_data} | {cheri_excep_cause_perm}   |`auth_cap` permission | not(auto_cap.<<w_perm>>)
-| all stores, all atomics           | {cheri_excep_mcause} | {cheri_excep_type_data} | {cheri_excep_cause_length} |`auth_cap` length     | any byte of access^!^ out of `auth_cap` bounds
+| all stores, all atomics           | {cheri_excep_mcause} | {cheri_excep_type_data} | {cheri_excep_cause_length} |`auth_cap` length     | any byte of access^1^ out of `auth_cap` bounds
 | capability stores, all atomics    |6                     | N/A                     | N/A                        |Misaligned store/AMO| Misaligned capability store or AMO
 |=========================================================================================
 
@@ -1119,7 +1120,7 @@ invalid addresses. Prior to writing these CSRs, implementations may convert an
 invalid address into some other invalid address that the register is capable of
 holding. However, these registers hold capabilities in {cheri_base_ext_name}
 and the bounds encoding depends on the address value, so implementations must
-not convert invalid addresses to other arbitrary invalid address in an
+not convert invalid addresses to other arbitrary invalid addresses in an
 unrestricted manner.
 The following procedure must be used instead when writing a capability A to
 these CSRs:

src/riscv-legacy-integration.adoc

@@ -142,7 +142,7 @@ address of the instruction following the jump (*pc* + 4) is written to an *x*
 register; that register's tag and capability metadata are zeroed.
 
 <<JAL>> and <<JALR>> cause CHERI exceptions when a minimum sized instruction
-at the target address are not within the bounds of the <<pcc>>. An
+at the target address is not within the bounds of the <<pcc>>. An
 instruction address misaligned exception is raised when the target address is
 misaligned.