Re: [PATCH] arm64: vdso: fix AArch32 compat init allocation leaks

[Will Deacon]

On Tue, Mar 24, 2026 at 11:09:12AM +0100, Thomas Weißschuh wrote:
> On Tue, Mar 24, 2026 at 09:59:15AM +0000, Will Deacon wrote:
> > On Mon, Mar 23, 2026 at 10:41:16PM +0100, Osama Abdelkader wrote:
> > > aarch32_alloc_vdso_pages() allocates the AA32 vdso pagelist, the compat
> > > sigpage, then the kuser vectors page. If aarch32_alloc_sigpage() or
> > > aarch32_alloc_kuser_vdso_page() fails, earlier allocations were not freed.
> > 
> > But why should they be freed? The vectors, sigpage and vdso are
> > independent from one another, so we can limp along with whatever we
> > managed to allocate. I'm not sure how far we'll get, mind, if single
> > page allocations are failing at initcall time...
> 
> In the core vDSO datastore we just panic() if the allocation fails.
> (See tip/timers/vdso for the currentl implementation)
> The same should work for the architecture-specific bits.

I think we should just leave the code as-is tbh unless there's an actual
issue here.

Will