This started simply to remove the two page[TARGET_PAGE_SIZE] instances. These turn into variable length arrays, when I start to allow the page size to vary for linux-user.
However, the first thing I noticed is that it is silly to write out target memory to the corefile page by page. As I started to clean that up, I noticed some actual errors in the writing of notes. Finally, we can stop creating local data structures to represent vmas and rely on the ones over in user-exec.c. r~ Richard Henderson (14): linux-user/elfload: Disable core dump if getrlimit fails linux-user/elfload: Merge init_note_info and fill_note_info linux-user/elfload: Tidy fill_note_info and struct elf_note_info linux-user/elfload: Stack allocate struct mm_struct linux-user/elfload: Latch errno before cleanup in elf_core_dump linux-user/elfload: Open core file after vma_init linux-user/elfload: Truncate core file on open linux-user/elfload: Lock cpu list and mmap during elf_core_dump linux-user/elfload: Size corefile before opening linux-user/elfload: Write corefile elf header in one block linux-user/elfload: Write process memory to core file in larger chunks linux-user/elfload: Simplify vma_dump_size linux-user/elfload: Rely on walk_memory_regions for vmas linux-user/elfload: Unprotect regions before core dump linux-user/elfload.c | 721 +++++++++++++------------------------------ 1 file changed, 221 insertions(+), 500 deletions(-) -- 2.34.1
