Skip to content

Commit

Permalink
[libunwind][NFC] Fix whitespace in comments
Browse files Browse the repository at this point in the history
  • Loading branch information
@ldionne
ldionne committed Aug 31, 2023
1 parent 2fad6e6 commit ba55034
Showing 1 changed file with 40 additions and 40 deletions.
80 changes: 40 additions & 40 deletions libunwind/include/mach-o/compact_unwind_encoding.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -108,7 +108,7 @@ enum {
// are encoded in the UNWIND_X86_EBP_FRAME_REGISTERS bits as five 3-bit entries.
// Each entry contains which register to restore.
// UNWIND_X86_MODE_STACK_IMMD:
// A "frameless" (EBP not used as frame pointer) function with a small
// A "frameless" (EBP not used as frame pointer) function with a small
// constant stack size. To return, a constant (encoded in the compact
// unwind encoding) is added to the ESP. Then the return is done by
// popping the stack into the pc.
Expand All @@ -119,16 +119,16 @@ enum {
// UNWIND_X86_FRAMELESS_STACK_REG_PERMUTATION contains which registers were
// saved and their order.
// UNWIND_X86_MODE_STACK_IND:
// A "frameless" (EBP not used as frame pointer) function large constant
// A "frameless" (EBP not used as frame pointer) function large constant
// stack size. This case is like the previous, except the stack size is too
// large to encode in the compact unwind encoding. Instead it requires that
// the function contains "subl $nnnnnnnn,ESP" in its prolog. The compact
// large to encode in the compact unwind encoding. Instead it requires that
// the function contains "subl $nnnnnnnn,ESP" in its prolog. The compact
// encoding contains the offset to the nnnnnnnn value in the function in
// UNWIND_X86_FRAMELESS_STACK_SIZE.
// UNWIND_X86_FRAMELESS_STACK_SIZE.
// UNWIND_X86_MODE_DWARF:
// No compact unwind encoding is available. Instead the low 24-bits of the
// compact encoding is the offset of the DWARF FDE in the __eh_frame section.
// This mode is never used in object files. It is only generated by the
// This mode is never used in object files. It is only generated by the
// linker in final linked images which have only DWARF unwind info for a
// function.
//
Expand Down Expand Up @@ -233,36 +233,36 @@ enum {
// For x86_64 there are four modes for the compact unwind encoding:
// UNWIND_X86_64_MODE_RBP_FRAME:
// RBP based frame where RBP is push on stack immediately after return address,
// then RSP is moved to RBP. Thus, to unwind RSP is restored with the current
// EPB value, then RBP is restored by popping off the stack, and the return
// then RSP is moved to RBP. Thus, to unwind RSP is restored with the current
// EPB value, then RBP is restored by popping off the stack, and the return
// is done by popping the stack once more into the pc.
// All non-volatile registers that need to be restored must have been saved
// in a small range in the stack that starts RBP-8 to RBP-2040. The offset/8
// in a small range in the stack that starts RBP-8 to RBP-2040. The offset/8
// is encoded in the UNWIND_X86_64_RBP_FRAME_OFFSET bits. The registers saved
// are encoded in the UNWIND_X86_64_RBP_FRAME_REGISTERS bits as five 3-bit entries.
// Each entry contains which register to restore.
// Each entry contains which register to restore.
// UNWIND_X86_64_MODE_STACK_IMMD:
// A "frameless" (RBP not used as frame pointer) function with a small
// constant stack size. To return, a constant (encoded in the compact
// unwind encoding) is added to the RSP. Then the return is done by
// A "frameless" (RBP not used as frame pointer) function with a small
// constant stack size. To return, a constant (encoded in the compact
// unwind encoding) is added to the RSP. Then the return is done by
// popping the stack into the pc.
// All non-volatile registers that need to be restored must have been saved
// on the stack immediately after the return address. The stack_size/8 is
// encoded in the UNWIND_X86_64_FRAMELESS_STACK_SIZE (max stack size is 2048).
// The number of registers saved is encoded in UNWIND_X86_64_FRAMELESS_STACK_REG_COUNT.
// UNWIND_X86_64_FRAMELESS_STACK_REG_PERMUTATION contains which registers were
// saved and their order.
// saved and their order.
// UNWIND_X86_64_MODE_STACK_IND:
// A "frameless" (RBP not used as frame pointer) function large constant
// A "frameless" (RBP not used as frame pointer) function large constant
// stack size. This case is like the previous, except the stack size is too
// large to encode in the compact unwind encoding. Instead it requires that
// the function contains "subq $nnnnnnnn,RSP" in its prolog. The compact
// large to encode in the compact unwind encoding. Instead it requires that
// the function contains "subq $nnnnnnnn,RSP" in its prolog. The compact
// encoding contains the offset to the nnnnnnnn value in the function in
// UNWIND_X86_64_FRAMELESS_STACK_SIZE.
// UNWIND_X86_64_FRAMELESS_STACK_SIZE.
// UNWIND_X86_64_MODE_DWARF:
// No compact unwind encoding is available. Instead the low 24-bits of the
// compact encoding is the offset of the DWARF FDE in the __eh_frame section.
// This mode is never used in object files. It is only generated by the
// This mode is never used in object files. It is only generated by the
// linker in final linked images which have only DWARF unwind info for a
// function.
//
Expand Down Expand Up @@ -307,20 +307,20 @@ enum {
// This is a standard arm64 prolog where FP/LR are immediately pushed on the
// stack, then SP is copied to FP. If there are any non-volatile registers
// saved, then are copied into the stack frame in pairs in a contiguous
// range right below the saved FP/LR pair. Any subset of the five X pairs
// range right below the saved FP/LR pair. Any subset of the five X pairs
// and four D pairs can be saved, but the memory layout must be in register
// number order.
// number order.
// UNWIND_ARM64_MODE_FRAMELESS:
// A "frameless" leaf function, where FP/LR are not saved. The return address
// A "frameless" leaf function, where FP/LR are not saved. The return address
// remains in LR throughout the function. If any non-volatile registers
// are saved, they must be pushed onto the stack before any stack space is
// allocated for local variables. The stack sized (including any saved
// non-volatile registers) divided by 16 is encoded in the bits
// non-volatile registers) divided by 16 is encoded in the bits
// UNWIND_ARM64_FRAMELESS_STACK_SIZE_MASK.
// UNWIND_ARM64_MODE_DWARF:
// No compact unwind encoding is available. Instead the low 24-bits of the
// compact encoding is the offset of the DWARF FDE in the __eh_frame section.
// This mode is never used in object files. It is only generated by the
// This mode is never used in object files. It is only generated by the
// linker in final linked images which have only DWARF unwind info for a
// function.
//
Expand All @@ -337,19 +337,19 @@ enum {

//
// A compiler can generated compact unwind information for a function by adding
// a "row" to the __LD,__compact_unwind section. This section has the
// S_ATTR_DEBUG bit set, so the section will be ignored by older linkers.
// It is removed by the new linker, so never ends up in final executables.
// This section is a table, initially with one row per function (that needs
// a "row" to the __LD,__compact_unwind section. This section has the
// S_ATTR_DEBUG bit set, so the section will be ignored by older linkers.
// It is removed by the new linker, so never ends up in final executables.
// This section is a table, initially with one row per function (that needs
// unwind info). The table columns and some conceptual entries are:
//
// range-start pointer to start of function/range
// range-length
// compact-unwind-encoding 32-bit encoding
// range-length
// compact-unwind-encoding 32-bit encoding
// personality-function or zero if no personality function
// lsda or zero if no LSDA data
//
// The length and encoding fields are 32-bits. The other are all pointer sized.
// The length and encoding fields are 32-bits. The other are all pointer sized.
//
// In x86_64 assembly, these entry would look like:
//
Expand All @@ -372,23 +372,23 @@ enum {
// .quad except_tab1
//
//
// Notes: There is no need for any labels in the the __compact_unwind section.
// The use of the .set directive is to force the evaluation of the
// Notes: There is no need for any labels in the the __compact_unwind section.
// The use of the .set directive is to force the evaluation of the
// range-length at assembly time, instead of generating relocations.
//
// To support future compiler optimizations where which non-volatile registers
// To support future compiler optimizations where which non-volatile registers
// are saved changes within a function (e.g. delay saving non-volatiles until
// necessary), there can by multiple lines in the __compact_unwind table for one
// function, each with a different (non-overlapping) range and each with
// different compact unwind encodings that correspond to the non-volatiles
// function, each with a different (non-overlapping) range and each with
// different compact unwind encodings that correspond to the non-volatiles
// saved at that range of the function.
//
// If a particular function is so wacky that there is no compact unwind way
// to encode it, then the compiler can emit traditional DWARF unwind info.
// to encode it, then the compiler can emit traditional DWARF unwind info.
// The runtime will use which ever is available.
//
// Runtime support for compact unwind encodings are only available on 10.6
// and later. So, the compiler should not generate it when targeting pre-10.6.
// Runtime support for compact unwind encodings are only available on 10.6
// and later. So, the compiler should not generate it when targeting pre-10.6.



Expand All @@ -402,7 +402,7 @@ enum {
//
// The __TEXT,__unwind_info section is laid out for an efficient two level lookup.
// The header of the section contains a coarse index that maps function address
// to the page (4096 byte block) containing the unwind info for that function.
// to the page (4096 byte block) containing the unwind info for that function.
//

#define UNWIND_SECTION_VERSION 1
Expand Down

0 comments on commit ba55034

Please sign in to comment.