From: Pasha Tatashin <pasha.tatashin@soleen.com>
To: Pratyush Yadav <pratyush@kernel.org>
Cc: akpm@linux-foundation.org, brauner@kernel.org, corbet@lwn.net,
graf@amazon.com, jgg@ziepe.ca, linux-kernel@vger.kernel.org,
linux-kselftest@vger.kernel.org, linux-mm@kvack.org,
masahiroy@kernel.org, ojeda@kernel.org, rdunlap@infradead.org,
rppt@kernel.org, tj@kernel.org, jasonmiu@google.com,
dmatlack@google.com, skhawaja@google.com
Subject: Re: [PATCH 1/2] liveupdate: kho: warn and fail on metadata or preserved memory in scratch area
Date: Wed, 15 Oct 2025 08:40:26 -0400 [thread overview]
Message-ID: <CA+CK2bD4z_j_jMhMtrvvQ5nOMEFT6yrRHk9rk7BEiQPSPCHiqA@mail.gmail.com> (raw)
In-Reply-To: <mafs0zf9sjrgc.fsf@kernel.org>
On Wed, Oct 15, 2025 at 8:10 AM Pratyush Yadav <pratyush@kernel.org> wrote:
>
> On Wed, Oct 15 2025, Pasha Tatashin wrote:
>
> > It is invalid for KHO metadata or preserved memory regions to be located
> > within the KHO scratch area, as this area is overwritten when the next
> > kernel is loaded, and used early in boot by the next kernel. This can
> > lead to memory corruption.
> >
> > Adds checks to kho_preserve_* and KHO's internal metadata allocators
> > (xa_load_or_alloc, new_chunk) to verify that the physical address of the
> > memory does not overlap with any defined scratch region. If an overlap
> > is detected, the operation will fail and a WARN_ON is triggered. To
> > avoid performance overhead in production kernels, these checks are
> > enabled only when CONFIG_KEXEC_HANDOVER_DEBUG is selected.
> >
> > Signed-off-by: Pasha Tatashin <pasha.tatashin@soleen.com>
> > ---
> > kernel/liveupdate/Kconfig | 15 ++++++++++
> > kernel/liveupdate/kexec_handover.c | 32 ++++++++++++++++++---
> > kernel/liveupdate/kexec_handover_debug.c | 18 ++++++++++++
> > kernel/liveupdate/kexec_handover_internal.h | 9 ++++++
> > 4 files changed, 70 insertions(+), 4 deletions(-)
> >
> > diff --git a/kernel/liveupdate/Kconfig b/kernel/liveupdate/Kconfig
> > index 522b9f74d605..d119f4f3f4b1 100644
> > --- a/kernel/liveupdate/Kconfig
> > +++ b/kernel/liveupdate/Kconfig
> > @@ -27,4 +27,19 @@ config KEXEC_HANDOVER_DEBUGFS
> > Also, enables inspecting the KHO fdt trees with the debugfs binary
> > blobs.
> >
> > +config KEXEC_HANDOVER_DEBUG
> > + bool "Enable Kexec Handover debug checks"
> > + depends on KEXEC_HANDOVER_DEBUGFS
>
> Why the dependency on debugfs? Why can't the debug checks be enabled
> independently?
Because there is one kexec_handover_debug.c file, that I thought would
make sense to use for both, but now thinking about this, perhaps we
should split the code: KEXEC_HANDOVER_DEBUGFS and
KEXEC_HANDOVER_DEBUG, and add two files:
kexec_handover_debugfs.c and kexec_handover_debug.c, this would avoid
ifdefs in .c.
>
> > + help
> > + This option enables extra sanity checks for the Kexec Handover
> > + subsystem.
> > +
> > + These checks verify that neither preserved memory regions nor KHO's
> > + internal metadata are allocated from within a KHO scratch area.
> > + An overlap can lead to memory corruption during a subsequent kexec
> > + operation.
>
> I don't think the checks that are done should be listed here since as
> soon as another check is added this list will become out of date.
I thought it could be expanded when new features are added, but I can
remove this description.
>
> > +
> > + If an overlap is detected, the kernel will print a warning and the
> > + offending operation will fail. This should only be enabled for
>
> This also describes the behaviour of the checks, which might change
> later. Maybe for some checks the operation won't fail? I suppose just
> leave it at "the kernel will print a warning"?
If it changes, and Kconfig should be updated as well.
>
> > + debugging purposes due to runtime overhead.
> > endmenu
> > diff --git a/kernel/liveupdate/kexec_handover.c b/kernel/liveupdate/kexec_handover.c
> > index 5da21f1510cc..ef1e6f7a234b 100644
> > --- a/kernel/liveupdate/kexec_handover.c
> > +++ b/kernel/liveupdate/kexec_handover.c
> > @@ -141,6 +141,11 @@ static void *xa_load_or_alloc(struct xarray *xa, unsigned long index, size_t sz)
> > if (!elm)
> > return ERR_PTR(-ENOMEM);
> >
> > + if (WARN_ON(kho_scratch_overlap(virt_to_phys(elm), sz))) {
> > + kfree(elm);
> > + return ERR_PTR(-EINVAL);
> > + }
> > +
> > res = xa_cmpxchg(xa, index, NULL, elm, GFP_KERNEL);
> > if (xa_is_err(res))
> > res = ERR_PTR(xa_err(res));
> > @@ -354,7 +359,13 @@ static struct khoser_mem_chunk *new_chunk(struct khoser_mem_chunk *cur_chunk,
> >
> > chunk = kzalloc(PAGE_SIZE, GFP_KERNEL);
> > if (!chunk)
> > - return NULL;
> > + return ERR_PTR(-ENOMEM);
> > +
> > + if (WARN_ON(kho_scratch_overlap(virt_to_phys(chunk), PAGE_SIZE))) {
> > + kfree(chunk);
> > + return ERR_PTR(-EINVAL);
> > + }
> > +
> > chunk->hdr.order = order;
> > if (cur_chunk)
> > KHOSER_STORE_PTR(cur_chunk->hdr.next, chunk);
> > @@ -379,14 +390,17 @@ static int kho_mem_serialize(struct kho_out *kho_out)
> > struct khoser_mem_chunk *chunk = NULL;
> > struct kho_mem_phys *physxa;
> > unsigned long order;
> > + int ret = -ENOMEM;
> >
> > xa_for_each(&kho_out->track.orders, order, physxa) {
> > struct kho_mem_phys_bits *bits;
> > unsigned long phys;
> >
> > chunk = new_chunk(chunk, order);
> > - if (!chunk)
> > + if (IS_ERR(chunk)) {
> > + ret = PTR_ERR(chunk);
> > goto err_free;
> > + }
> >
> > if (!first_chunk)
> > first_chunk = chunk;
> > @@ -396,8 +410,10 @@ static int kho_mem_serialize(struct kho_out *kho_out)
> >
> > if (chunk->hdr.num_elms == ARRAY_SIZE(chunk->bitmaps)) {
> > chunk = new_chunk(chunk, order);
> > - if (!chunk)
> > + if (IS_ERR(chunk)) {
> > + ret = PTR_ERR(chunk);
> > goto err_free;
> > + }
> > }
> >
> > elm = &chunk->bitmaps[chunk->hdr.num_elms];
> > @@ -414,7 +430,7 @@ static int kho_mem_serialize(struct kho_out *kho_out)
> >
> > err_free:
> > kho_mem_ser_free(first_chunk);
> > - return -ENOMEM;
> > + return ret;
> > }
> >
> > static void __init deserialize_bitmap(unsigned int order,
> > @@ -737,6 +753,9 @@ int kho_preserve_folio(struct folio *folio)
> > const unsigned int order = folio_order(folio);
> > struct kho_mem_track *track = &kho_out.track;
> >
> > + if (WARN_ON(kho_scratch_overlap(pfn << PAGE_SHIFT, PAGE_SIZE << order)))
> > + return -EINVAL;
> > +
> > return __kho_preserve_order(track, pfn, order);
> > }
> > EXPORT_SYMBOL_GPL(kho_preserve_folio);
> > @@ -784,6 +803,11 @@ int kho_preserve_pages(struct page *page, unsigned int nr_pages)
> > unsigned long failed_pfn = 0;
> > int err = 0;
> >
> > + if (WARN_ON(kho_scratch_overlap(start_pfn << PAGE_SHIFT,
> > + nr_pages << PAGE_SHIFT))) {
> > + return -EINVAL;
> > + }
> > +
> > while (pfn < end_pfn) {
> > const unsigned int order =
> > min(count_trailing_zeros(pfn), ilog2(end_pfn - pfn));
> > diff --git a/kernel/liveupdate/kexec_handover_debug.c b/kernel/liveupdate/kexec_handover_debug.c
> > index eb47f000887d..294d1d290142 100644
> > --- a/kernel/liveupdate/kexec_handover_debug.c
> > +++ b/kernel/liveupdate/kexec_handover_debug.c
> > @@ -214,3 +214,21 @@ __init int kho_debugfs_init(void)
> > return -ENOENT;
> > return 0;
> > }
> > +
> > +#ifdef CONFIG_KEXEC_HANDOVER_DEBUG
> > +bool kho_scratch_overlap(phys_addr_t phys, size_t size)
> > +{
> > + phys_addr_t scratch_start, scratch_end;
> > + unsigned int i;
> > +
> > + for (i = 0; i < kho_scratch_cnt; i++) {
> > + scratch_start = kho_scratch[i].addr;
> > + scratch_end = kho_scratch[i].addr + kho_scratch[i].size - 1;
>
> Nit: wouldn't it be a tad bit simpler to do
>
> scratch_end = kho_scratch[i].addr + kho_scratch[i].size;
>
> > +
> > + if (phys <= scratch_end && (phys + size) > scratch_start)
>
> and here
>
> if (phys < scratch_end && (phys + size) > scratch_start)
>
> At least I find it slightly easier to understand, though I don't think
> it makes too much of a difference so either way is fine.
>
> > + return true;
> > + }
> > +
> > + return false;
> > +}
> > +#endif /* CONFIG_KEXEC_HANDOVER_DEBUG */
> > diff --git a/kernel/liveupdate/kexec_handover_internal.h b/kernel/liveupdate/kexec_handover_internal.h
> > index b3fc1957affa..92798346fa5a 100644
> > --- a/kernel/liveupdate/kexec_handover_internal.h
> > +++ b/kernel/liveupdate/kexec_handover_internal.h
> > @@ -44,4 +44,13 @@ static inline void kho_debugfs_fdt_remove(struct kho_debugfs *dbg,
> > void *fdt) { }
> > #endif /* CONFIG_KEXEC_HANDOVER_DEBUGFS */
> >
> > +#ifdef CONFIG_KEXEC_HANDOVER_DEBUG
> > +bool kho_scratch_overlap(phys_addr_t phys, size_t size);
> > +#else
> > +static inline bool kho_scratch_overlap(phys_addr_t phys, size_t size)
> > +{
> > + return false;
> > +}
> > +#endif /* CONFIG_KEXEC_HANDOVER_DEBUG */
> > +
> > #endif /* LINUX_KEXEC_HANDOVER_INTERNAL_H */
>
> --
> Regards,
> Pratyush Yadav
next prev parent reply other threads:[~2025-10-15 12:41 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-10-15 5:31 [PATCH 0/2] KHO: Fix metadata allocation " Pasha Tatashin
2025-10-15 5:31 ` [PATCH 1/2] liveupdate: kho: warn and fail on metadata or preserved memory " Pasha Tatashin
2025-10-15 8:21 ` Mike Rapoport
2025-10-15 12:36 ` Pasha Tatashin
2025-10-16 17:23 ` Mike Rapoport
2025-10-18 15:31 ` Pasha Tatashin
2025-10-18 15:28 ` Pasha Tatashin
2025-10-15 12:10 ` Pratyush Yadav
2025-10-15 12:40 ` Pasha Tatashin [this message]
2025-10-15 13:11 ` Pratyush Yadav
2025-10-15 5:31 ` [PATCH 2/2] liveupdate: kho: allocate metadata directly from the buddy allocator Pasha Tatashin
2025-10-15 8:37 ` Mike Rapoport
2025-10-15 12:46 ` Pasha Tatashin
2025-10-15 13:05 ` Pratyush Yadav
2025-10-15 14:19 ` Pasha Tatashin
2025-10-15 14:36 ` Alexander Potapenko
2025-10-24 13:25 ` Jason Gunthorpe
2025-10-24 13:57 ` Pasha Tatashin
2025-10-24 14:20 ` Jason Gunthorpe
2025-10-24 14:36 ` Pasha Tatashin
2025-10-24 14:55 ` Jason Gunthorpe
2025-10-24 15:06 ` Pasha Tatashin
2025-10-15 14:22 ` Pasha Tatashin
2025-10-24 13:21 ` Jason Gunthorpe
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CA+CK2bD4z_j_jMhMtrvvQ5nOMEFT6yrRHk9rk7BEiQPSPCHiqA@mail.gmail.com \
--to=pasha.tatashin@soleen.com \
--cc=akpm@linux-foundation.org \
--cc=brauner@kernel.org \
--cc=corbet@lwn.net \
--cc=dmatlack@google.com \
--cc=graf@amazon.com \
--cc=jasonmiu@google.com \
--cc=jgg@ziepe.ca \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=masahiroy@kernel.org \
--cc=ojeda@kernel.org \
--cc=pratyush@kernel.org \
--cc=rdunlap@infradead.org \
--cc=rppt@kernel.org \
--cc=skhawaja@google.com \
--cc=tj@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox