Change Details

This implements https://sourceware.org/bugzilla/show_bug.cgi?id=26404#c6 --overwrite-script specifies a script which contains several `SECTIONS` commands. Each `SECTIONS` specifies exactly one output section description. The initial syntax is intentionally narrow. Symbol assignments cannot be used. Not supporting multiple output section descriptions in one `SECTIONS` makes it clear the sections don't have order guarantee. The semantics are similar to `INSERT [BEFORE|AFTER]`: LinkerScript::hasSectionsCommand is not set. --overwrite-script can be more convenient without a main script because --overwrite-script does not need a random anchor section used by `INSERT`. This feature is versatile. To list a few usage: * Use `section : { KEEP(...) }` to retain input sections under GC * Define encapsulation symbols (start/end) for an output section * Use `section : ALIGN(...) : { ... }` to overalign an output section (similar to ld64 `-sectalign`)

This implements https://sourceware.org/bugzilla/show_bug.cgi?id=26404#c6 --overwrite-section-script specifies a script which contains several `SECTIONS` commands. Each `SECTIONS` specifies exactly one output section description. The initial syntax is intentionally narrow to facilitate backward compatible extensions with backward. Symbol assignments cannot be used. Not supporting multiple output section descriptions in one `SECTIONS` makes it clear the sections don't have order guarantee. The semantics are similar to `INSERT [BEFORE|AFTER]` in that LinkerScript::hasSectionsCommand is not set, so default linker script rules still apply. In the absence of a main linker script, --overwrite-script can be more convenient than `INSERT` because --overwrite-script does not need a random anchor section used by `INSERT`. This feature is versatile. To list a few usage: * Use `section : { KEEP(...) }` to retain input sections under GC * Define encapsulation symbols (start/end) for an output section * Use `section : ALIGN(...) : { ... }` to overalign an output section (similar to ld64 `-sectalign`) When an output section is specified by both `--overwrite-section-script` and `INSERT`, `INSERT` is processed after overwrite sections. To make this work, this patch changes `InsertCommand` to use name based matching instead of pointer based matching. This causes a difference when `INSERT` moves one output section more than once. This should not be used in practice (seems that in GNU ld the output sections may just disappear)

This implements https://sourceware.org/bugzilla/show_bug.cgi?id=26404#c6 --overwrite-section-script specifies a script which contains several `SECTIONS` commands. Each `SECTIONS` specifies exactly one output section description. The initial syntax is intentionally narrow. Symbol assignments cannot be used. to facilitate backward compatible Not supporting multiple output section descriptions in one `SECTIONS` makes itextensions with backward. Symbol assignments cannot be used. Not supporting multiple output section descriptions in one `SECTIONS` makes it clear the sections don't have order guarantee. The semantics are similar to `INSERT [BEFORE|AFTER]`: LinkerScript::hasSectionsCommand in that LinkerScript::hasSectionsCommand is not set, so default linker script rules still apply. In the absence of a main linker script, --overwrite-script can be is not set. --overwrite-script can be more convenient without a main scriptmore convenient than `INSERT` because --overwrite-script does not need a random because --overwrite-script does not need a random anchor section used by `INSERT`. This feature is versatile. To list a few usage: * Use `section : { KEEP(...) }` to retain input sections under GC * Define encapsulation symbols (start/end) for an output section * Use `section : ALIGN(...) : { ... }` to overalign an output section (similar to ld64 `-sectalign`) When an output section is specified by both `--overwrite-section-script` and `INSERT`, `INSERT` is processed after overwrite sections. To make this work, this patch changes `InsertCommand` to use name based matching instead of pointer based matching. This causes a difference when `INSERT` moves one output section more than once. This should not be used in practice (seems that in GNU ld the output sections may just disappear)