Ignore too large stack in case of dsl_deadlist_merge

[oshogbo]

Motivation and Context

In the case of a regular compilation, the compiler raises a warning for a dsl_deadlist_merge function. However, in debug build this can generate an error.

For this single function, let's ignore the limit of stack size.

Signed-off-by: Mariusz Zaborski mariusz.zaborski@klarasystems.com
Sponsored-by: Klara, Inc.
Sponsored-by: Wasabi Technology, Inc.

Description

For this single function override the compiler option of stack size.

How Has This Been Tested?

I have built a project using gcc and clang.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Performance enhancement (non-breaking change which improves efficiency)
• Code cleanup (non-breaking change which makes code smaller or more readable)
• Breaking change (fix or feature that would cause existing functionality to change)
• Library ABI change (libzfs, libzfs_core, libnvpair, libuutil and libzfsbootenv)
• Documentation (a change to man pages or other documentation)

Checklist:

• My code follows the OpenZFS code style requirements.
• I have updated the documentation accordingly.
• I have read the contributing document.
• I have added tests to cover my changes.
• I have run the ZFS Test Suite with this change applied.
• All commit messages are properly formatted and contain Signed-off-by.

[amotin]

Does it tell how big is the frame? Can it be fixed instead by some noinline's?

[oshogbo]

In debug:

error: the frame size of 1040 bytes is larger than 1024 bytes

[amotin]

error: the frame size of 1040 bytes is larger than 1024 bytes

That is quite a lot for kernel. Would be good to make sure we've done everything we could to reduce it.

[behlendorf]

It looks like the two zap_attribute_t locals za, and pza are responsible for over half of the usage. Let's just relocate them to the heap.

[ryao]

It would be more efficient to do a single allocation like this:

za = kmem_alloc(sizeof (*za) * 2, KM_SLEEP);
pza = &za[1];

Then we would free the memory by calling kmem_free(za, sizeof (*za) * 2); at the end of the function.

That said, I do not feel strongly about this, so I am fine with merging this either way. This is purely a suggestion for a way to minimize the runtime overhead of using dynamic memory to reduce stack space usage that stood out to me.