[Dwarf-discuss] Re: LSB DWARF Chapter

[Ron 603-884-2088]

Following are my notes from reviewing what I think is the
only DWARF-related part of the ELF component of LSB document.

Ron Brender

-----------------------------------------------------------------

Section 7.1, Table 7-1: It seems odd to see 0xff as the value of
a 4-bit field (the value format nibble).

If the intent that only the combination of 0x0f in the format
nibble and 0xf0 in the "application" nibble (see Table 7-2),
then this needs to be called out much more explicitly.

Otherwise, correct this typo to read 0x0f.

Section 7.1, Table 7-2: The last line of this table makes no
sense. First, its value duplicates a value already used for
DW_EH_PE_datarel. Second, what does "value is aligned" mean?
Aligned with what?

Section 7.2, and later: As Dave Anderson suggested, the
references to DWARF Debugging Information Format might better
1) refer to the FSG's own website rather than eagercon.com
and 2) might better reference DWARF V3 rather than V2.

Section 7.2, DW_CFA_GNU_args_size: The Meaning column talks
about DW_CFA_def_cfa_offset_sf where it seems DW_CFA_GNU_args_size
must be intended.

Section 7.3: It would help to clarify up front that this
design is "based on" but extends DWARF, using non-DWARF
section names.

Section 7.3, description of Length: insert missing "in" following
"then the length is contained".

Section 7.3.1, description of EH Data: The description is incomplete.
Replace the two occurences of "..." and tell us the rest of the
story.

Section 7.3.1.1: What an odd mixture of upper case and lower case.
I presume that 'z' and "eh" must be only lower case and 'L', 'P'
and 'R' must be only upper case -- perhaps you could say so more
explicitly.

In the description of 'z', insert "Augmentation Length and" in front
of "Augmentation Data". At least I presume it is not meaningful to
have any Data without a preceeding Length.

It appears that 'L', 'P' or 'R' may be present if and only if 'z'
is present. It appears that the order of 'L', 'P' and 'R' is
unimportant. It appears that "eh" may occur anywhere, interspersed
with 'L', 'P', and 'R, (including before or after) except at the
beginning unless none of the others are present (ie, it can occur
by itself). If any of these statements are not the intent, then
please clarify (actually, even if they are the intent, more
explicit statement would be good).

Section 7.4: The second paragraph refers to the "DWARF Exception
Header Encoding". DWARF has no such encoding/section/concept, so
it is misleading to describe or name the .eh_frame_hdr in this
way. It is purely a ELF adaptation/modification of the base DWARF
call frame representation and concepts.