Re: [PATCH v3 03/14] dump: Convert GuestPhysBlock iterators and use the filter functions

[Janosch Frank]

On 7/21/22 16:36, Marc-André Lureau wrote:
[..]
> > diff --git a/include/sysemu/dump.h b/include/sysemu/dump.h
> > index 2b39abeeae..7025e50682 100644
> > --- a/include/sysemu/dump.h
> > +++ b/include/sysemu/dump.h
> > @@ -166,11 +166,10 @@ typedef struct DumpState {
> >       hwaddr memory_offset;
> >       int fd;
> >
> > -    GuestPhysBlock *next_block;
> > -    ram_addr_t start;
> > -    bool has_filter;
> > -    int64_t begin;
> > -    int64_t length;
> > +    /* Guest memory related data */
> > +    bool has_filter;           /* Are we dumping parts of the memory? */
> > +    int64_t begin;             /* Start address of the chunk we want to dump */
> > +    int64_t length;            /* Length of the dump we want to dump */
> >
> >       uint8_t *note_buf;          /* buffer for notes */
> >       size_t note_buf_offset;     /* the writing place in note_buf */
> > --
> > 2.34.1
>
> My suggestion in v2 review was to introduce each function &
> refactoring independently, if possible. And it looks like the
> validate_start_block() change could be a 3rd patch too.

Alright, I just squashed and split this into 5 patches:
 * Introducing the 2 new functions
 * Converting dump_iterate and removing get_next_block
 * get_start_block -> validate_start_block
 * Removal of next_block and start from DumpState, last user was 
get_start_block
 * Re-work of dump_calculate_size


I don't think we can easily adapt to dump_get_memblock_size() and 
dump_get_memblock_start() independently.

I'll also move the DumpState comment hunk to the removal of start and 
next_block tomorrow.