[PATCH v5 03/23] efi_loader: Add comments where incorrect addresses are used
Simon Glass
sjg at chromium.org
Thu Dec 12 14:43:52 CET 2024
Hi Ilias,
On Wed, 11 Dec 2024 at 23:15, Ilias Apalodimas
<ilias.apalodimas at linaro.org> wrote:
>
> On Wed, 11 Dec 2024 at 17:58, Simon Glass <sjg at chromium.org> wrote:
> >
> > Hi Ilias,
> >
> > On Wed, 11 Dec 2024 at 07:48, Ilias Apalodimas
> > <ilias.apalodimas at linaro.org> wrote:
> > >
> > > On Wed, 11 Dec 2024 at 15:54, Simon Glass <sjg at chromium.org> wrote:
> > > >
> > > > Some functions are passing addresses instead of pointers to the
> > > > efi_add_memory_map() function. This confusion is understandable since
> > > > the function arguments indicate an address.
> > > >
> > > > Make a note of the 8 places where there are problems, which would break
> > > > usage in sandbox tests.
> > > >
> > > > Future work will resolve these problems.
> > >
> > > NAK, this serves no purpose whatsoever. Just send a patch with
> > > whatever changes you think are needed
> >
> > Please see below
> >
> > >
> > > Thanks
> > > /Ilias
> > > >
> > > > Signed-off-by: Simon Glass <sjg at chromium.org>
> > > > ---
> > > > A request was made in v4 to drop this patch, but in the same thread, a
> > > > query was sent about why we need these addresses, so at least until the
> > > > series is about to be applied, I've left it in, so we can have that
> > > > discussion in one place and build a shared understanding of why these
> > > > addresses are needed, instead of peppering the rest of the series with
> > > > similar questions.
> >
> > I'm quite happy to drop this patch. But I want to first make sure that
> > you understand that these bugs exist today in the code, albeit they
> > only matter on sandbox.
>
> Thei weird thing is sandbox works perfectly fine in EFI, passes all
> internal tests, EFI SCT tests, loads binaries etc. Any idea why that
> works?
>From what I can tell, the code mentioned is not tested by sandbox yet.
It would just segfault. We do have a lot of self tests, but my
bootflow_efi should allow testing of more features over time. I
noticed the runtime-code issue when looking at an EFI log.
BTW how are you running that? When I tried it on x86_64 it just
crashed. Are you using the sandbox build on an Arm64 machine?
>
> > You will notice that my future patches don't
> > change the code mentioned in this file. The bugs are simply resolved
> > by reworking the EFI-loader memory map to use addresses, instead of
> > pointers cast to u64
> >
> > So can you please confirm this makes sense? (if not, let's discuss
> > it). Then I can drop this patch and the later one.
>
> The ptrs to u64 doesn't make too much sense tbh. I do understand how
> sandbox works, but EFI deals with physical addresses which are also
> mapped 1:1
Yes, and that's why (in efi_boottime) the conversion is done in one
place (from u64 pointers/addresses to U-Boot addresses). It simplifies
the code and avoids the kind of bugs here.
Regards,
Simon
>
> Thanks
> /Ilias
> >
> > > >
> > > > Link: https://lore.kernel.org/u-boot/CAC_iWjLex4Z41CDpx4u07XeCYgjB7H+Ynd8aQYL-wruH7yRmQA@mail.gmail.com/
> > > >
> > > > (no changes since v2)
> > > >
> > > > Changes in v2:
> > > > - Add a new patch with comments where incorrect addresses are used
> > > >
> > > > lib/efi_loader/efi_acpi.c | 2 ++
> > > > lib/efi_loader/efi_bootmgr.c | 5 +++++
> > > > lib/efi_loader/efi_helper.c | 1 +
> > > > lib/efi_loader/efi_smbios.c | 6 +++++-
> > > > lib/efi_loader/efi_var_mem.c | 2 ++
> > > > 5 files changed, 15 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/lib/efi_loader/efi_acpi.c b/lib/efi_loader/efi_acpi.c
> > > > index 67bd7f8ca24..ffb360d461b 100644
> > > > --- a/lib/efi_loader/efi_acpi.c
> > > > +++ b/lib/efi_loader/efi_acpi.c
> > > > @@ -28,12 +28,14 @@ efi_status_t efi_acpi_register(void)
> > > > /* Mark space used for tables */
> > > > start = ALIGN_DOWN(gd->arch.table_start, EFI_PAGE_MASK);
> > > > end = ALIGN(gd->arch.table_end, EFI_PAGE_MASK);
> > > > + /* TODO(sjg): This should use (ulong)map_sysmem(start, ...) */
> > > > ret = efi_add_memory_map(start, end - start, EFI_ACPI_RECLAIM_MEMORY);
> > > > if (ret != EFI_SUCCESS)
> > > > return ret;
> > > > if (gd->arch.table_start_high) {
> > > > start = ALIGN_DOWN(gd->arch.table_start_high, EFI_PAGE_MASK);
> > > > end = ALIGN(gd->arch.table_end_high, EFI_PAGE_MASK);
> > > > + /* TODO(sjg): This should use (ulong)map_sysmem(start, ...) */
> > > > ret = efi_add_memory_map(start, end - start,
> > > > EFI_ACPI_RECLAIM_MEMORY);
> > > > if (ret != EFI_SUCCESS)
> > > > diff --git a/lib/efi_loader/efi_bootmgr.c b/lib/efi_loader/efi_bootmgr.c
> > > > index 8c51a6ef2ed..bb635d77b53 100644
> > > > --- a/lib/efi_loader/efi_bootmgr.c
> > > > +++ b/lib/efi_loader/efi_bootmgr.c
> > > > @@ -365,6 +365,8 @@ static efi_status_t prepare_loaded_image(u16 *label, ulong addr, ulong size,
> > > > * TODO: expose the ramdisk to OS.
> > > > * Need to pass the ramdisk information by the architecture-specific
> > > > * methods such as 'pmem' device-tree node.
> > > > + *
> > > > + * TODO(sjg): This should use (ulong)map_sysmem(addr)
> > > > */
> > > > ret = efi_add_memory_map(addr, size, EFI_RESERVED_MEMORY_TYPE);
> > > > if (ret != EFI_SUCCESS) {
> > > > @@ -399,6 +401,7 @@ static efi_status_t efi_bootmgr_release_uridp(struct uridp_context *ctx)
> > > >
> > > > /* cleanup for iso or img image */
> > > > if (ctx->ramdisk_blk_dev) {
> > > > + /* TODO(sjg): This should use (ulong)map_sysmem(...) */
> > > > ret = efi_add_memory_map(ctx->image_addr, ctx->image_size,
> > > > EFI_CONVENTIONAL_MEMORY);
> > > > if (ret != EFI_SUCCESS)
> > > > @@ -506,6 +509,8 @@ static efi_status_t try_load_from_uri_path(struct efi_device_path_uri *uridp,
> > > >
> > > > source_buffer = NULL;
> > > > source_size = 0;
> > > > +
> > > > + /* TODO(sjg): This does not work on sandbox */
> > > > } else if (efi_check_pe((void *)image_addr, image_size, NULL) == EFI_SUCCESS) {
> > > > /*
> > > > * loaded_dp must exist until efi application returns,
> > > > diff --git a/lib/efi_loader/efi_helper.c b/lib/efi_loader/efi_helper.c
> > > > index bf96f61d3d0..fd89ef6036f 100644
> > > > --- a/lib/efi_loader/efi_helper.c
> > > > +++ b/lib/efi_loader/efi_helper.c
> > > > @@ -486,6 +486,7 @@ static efi_status_t copy_fdt(void **fdtp)
> > > > log_err("Failed to reserve space for FDT\n");
> > > > goto done;
> > > > }
> > > > + /* TODO(sjg): This does not work on sandbox */
> > > > new_fdt = (void *)(uintptr_t)new_fdt_addr;
> > > > memcpy(new_fdt, fdt, fdt_totalsize(fdt));
> > > > fdt_set_totalsize(new_fdt, fdt_size);
> > > > diff --git a/lib/efi_loader/efi_smbios.c b/lib/efi_loader/efi_smbios.c
> > > > index 8d2ef6deb51..02d4a5dd045 100644
> > > > --- a/lib/efi_loader/efi_smbios.c
> > > > +++ b/lib/efi_loader/efi_smbios.c
> > > > @@ -40,7 +40,11 @@ efi_status_t efi_smbios_register(void)
> > > > return EFI_NOT_FOUND;
> > > > }
> > > >
> > > > - /* Mark space used for tables */
> > > > + /*
> > > > + * Mark space used for tables/
> > > > + *
> > > > + * TODO(sjg): This should use (ulong)map_sysmem(addr, ...)
> > > > + */
> > > > ret = efi_add_memory_map(addr, TABLE_SIZE, EFI_RUNTIME_SERVICES_DATA);
> > > > if (ret)
> > > > return ret;
> > > > diff --git a/lib/efi_loader/efi_var_mem.c b/lib/efi_loader/efi_var_mem.c
> > > > index 139e16aad7c..5c69a1e0f3e 100644
> > > > --- a/lib/efi_loader/efi_var_mem.c
> > > > +++ b/lib/efi_loader/efi_var_mem.c
> > > > @@ -226,6 +226,8 @@ efi_status_t efi_var_mem_init(void)
> > > > &memory);
> > > > if (ret != EFI_SUCCESS)
> > > > return ret;
> > > > +
> > > > + /* TODO(sjg): This does not work on sandbox */
> > > > efi_var_buf = (struct efi_var_file *)(uintptr_t)memory;
> > > > memset(efi_var_buf, 0, EFI_VAR_BUF_SIZE);
> > > > efi_var_buf->magic = EFI_VAR_FILE_MAGIC;
> > > > --
> > > > 2.34.1
> > > >
More information about the U-Boot
mailing list