[PATCH] malloc: Remove dumped heap support

[Sam James]

Adhemerval Zanella Netto <adhemerval.zanella@linaro.org> writes:

> On 16/09/25 15:39, Sam James wrote:
>> Adhemerval Zanella Netto <adhemerval.zanella@linaro.org> writes:
>> 
>>> On 30/07/25 06:33, Wilco Dijkstra wrote:
>>>> Hi Florian,
>>>>
>>>>>> Remove support for obsolete dumped heaps.  Dumping heaps was discontinued
>>>>>> 8 years ago, however loading a dumped heap is still supported. This blocks
>>>>>> changes and improvements of the malloc data structures - hence it is time
>>>>>> to remove this.  Ancient binaries that still call malloc_set_state will now
>>>>>> get the -1 error code.  Update tst-mallocstate.c to just check for this.
>>>>>
>>>>> What are the improvements blocked by the heap rewriting?
>>>>
>>>> Basically anything that changes the block layout and bit allocation in memory
>>>> is blocked by it. Even my minor change to mmap [1] was problematic since it
>>>> marks dumped heap blocks as mmaps and reuses the various mmap calls.
>>>>
>>>> There is also a fundamental issue with the structure of the code - including
>>>> malloc.c repeatedly to be used in different contexts is a terrible design.
>>>> Getting rid of anything we don't strictly need that relies on malloc internals
>>>> makes it easier to improve malloc.
>>>>
>>>> Cheers,
>>>> Wilco
>>>>
>>>> [1] https://sourceware.org/pipermail/libc-alpha/2025-July/169025.html
>>>
>>> So I take that there is no blocker for this change, I need it only requires
>>> a NEWS entry on 'Deprecated and removed features'.
>>>
>>> Reviewed-by: Adhemerval Zanella  <adhemerval.zanella@linaro.org>
>> 
>> I'm afraid I object until the author replies wrt Emacs, and ideally
>> seeks some input from the Emacs ML.
>> 
>> (I'm not against the change by itself and I don't think we need to
>> support that indefinitely, I just don't think we can do it willynilly.)
>
> Right, but from Florian's comment the emacs configure checks for 
> Doug Lea malloc style will always return false.

Right, this is fine.

> And afaik the 
> heap dump support is an startup optimization, so not really
> required to run old emacs binaries.

This is the part I'm not clear on. If it is just an optimisation, then
indeed, this is fine. My understanding (I'm not certain) is that it's
needed to actually run old binaries of a certain vintage.

(I am really not saying we have to support those forever, to be clear. I
just want us to make that decision being very clear on the facts, and
ideally informing the relevant community first.)