u-boot.lists.denx.de archive mirror
 help / color / mirror / Atom feed
From: AKASHI Takahiro <takahiro.akashi@linaro.org>
To: Simon Glass <sjg@chromium.org>
Cc: Heinrich Schuchardt <xypron.glpk@gmx.de>,
	U-Boot Mailing List <u-boot@lists.denx.de>,
	Ilias Apalodimas <ilias.apalodimas@linaro.org>,
	Bin Meng <bmeng.cn@gmail.com>, Tom Rini <trini@konsulko.com>,
	Christian Melki <christian.melki@t2data.com>,
	Alexander Graf <agraf@csgraf.de>
Subject: Re: [PATCH 14/35] efi: Locate all block devices in the app
Date: Fri, 10 Sep 2021 09:50:46 +0900	[thread overview]
Message-ID: <20210910005046.GA42777@laputa> (raw)
In-Reply-To: <CAPnjgZ0jFWfaq4FtLCGMuDuBbs4a23TZr50M5yrXVFX5X6nCOA@mail.gmail.com>

On Thu, Sep 09, 2021 at 02:09:12PM -0600, Simon Glass wrote:
> Hi Takahiro,
> 
> On Wed, 8 Sept 2021 at 19:11, AKASHI Takahiro
> <takahiro.akashi@linaro.org> wrote:
> >
> > Hi Simon,
> >
> > On Wed, Sep 08, 2021 at 08:14:30PM +0200, Heinrich Schuchardt wrote:
> > > On 9/8/21 3:33 PM, Simon Glass wrote:
> > > > When starting the app, locate all block devices and make them available
> > > > to U-Boot. This allows listing partitions and accessing files in
> > > > filesystems.
> > > >
> > > > EFI also has the concept of 'disks', meaning boot media. For now, this
> > > > is not obviously useful in U-Boot, but add code to at least locate these.
> > > > This can be expanded later as needed.
> > >
> > > UEFI firmware handles with the EFI_BLOCK_IO_PROTOCOL for raw access to
> > > disks and partitions. It further provides the
> > > EFI_SIMPLE_FILE_SYSTEM_PROTOCOL on these handles to access files on
> > > formatted media.
> >
> > Do you want to implement "efifs" as a U-Boot's file system on top of
> > SIMPLE_FILE_SYSTEM_PROTO?
> > Just kidding.
> 
> Eek.
> 
> >
> > I have one concern:
> >
> > > >
> > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > ---
> > > >
> > > >   include/efi.h     |  15 ++++++
> > > >   include/efi_api.h |  15 ++++++
> > > >   lib/efi/efi_app.c | 119 ++++++++++++++++++++++++++++++++++++++++++++++
> > > >   3 files changed, 149 insertions(+)
> > > >
> > > > diff --git a/include/efi.h b/include/efi.h
> > > > index 0ec5913ddd1..c0fddf7f6cd 100644
> > > > --- a/include/efi.h
> > > > +++ b/include/efi.h
> > > > @@ -529,4 +529,19 @@ void efi_putc(struct efi_priv *priv, const char ch);
> > > >    */
> > > >   int efi_info_get(enum efi_entry_t type, void **datap, int *sizep);
> > > >
> > > > +/**
> > > > + * efi_bind_block() - bind a new block device to an EFI device
> > > > + *
> > > > + * Binds a new top-level EFI_MEDIA device as well as a child block device so
> > > > + * that the block device can be accessed in U-Boot.
> > > > + *
> > > > + * The device can then be accessed using 'part list efi 0', 'fat ls efi 0:1',
> > > > + * for example, just like any other interface type.
> > > > + *
> > > > + * @handle: handle of the controller on which this driver is installed
> > > > + * @blkio: block io protocol proxied by this driver
> > > > + * @return 0 if OK, -ve on error
> > > > + */
> > > > +int efi_bind_block(efi_handle_t handle, struct efi_block_io *blkio);
> > > > +
> > > >   #endif /* _LINUX_EFI_H */
> > > > diff --git a/include/efi_api.h b/include/efi_api.h
> > > > index c8f959bb720..0e88b3e5dbe 100644
> > > > --- a/include/efi_api.h
> > > > +++ b/include/efi_api.h
> > > > @@ -1994,4 +1994,19 @@ struct efi_firmware_management_protocol {
> > > >                     const u16 *package_version_name);
> > > >   };
> > > >
> > > > +#define EFI_DISK_IO_PROTOCOL_GUID  \
> > > > +   EFI_GUID(0xce345171, 0xba0b, 0x11d2, 0x8e, 0x4f, \
> > > > +            0x00, 0xa0, 0xc9, 0x69, 0x72, 0x3b)
> > > > +
> > > > +struct efi_disk {
> > > > +   u64 revision;
> > > > +   efi_status_t (EFIAPI *read_disk)(struct efi_disk *this, u32 media_id,
> > > > +                                    u64 offset, efi_uintn_t buffer_size,
> > > > +                                    void *buffer);
> > > > +
> > > > +   efi_status_t (EFIAPI *write_disk)(struct efi_disk *this, u32 media_id,
> > > > +                                     u64 offset, efi_uintn_t buffer_size,
> > > > +                                     void *buffer);
> > > > +};
> > > > +
> > > >   #endif
> > > > diff --git a/lib/efi/efi_app.c b/lib/efi/efi_app.c
> > > > index f61665686c5..9ba48517422 100644
> > > > --- a/lib/efi/efi_app.c
> > > > +++ b/lib/efi/efi_app.c
> > > > @@ -21,6 +21,9 @@
> > > >   #include <efi.h>
> > > >   #include <efi_api.h>
> > > >   #include <sysreset.h>
> > > > +#include <dm/device-internal.h>
> > > > +#include <dm/lists.h>
> > > > +#include <dm/root.h>
> > > >
> > > >   DECLARE_GLOBAL_DATA_PTR;
> > > >
> > > > @@ -46,6 +49,33 @@ int efi_info_get(enum efi_entry_t type, void **datap, int *sizep)
> > > >     return -ENOSYS;
> > > >   }
> > > >
> > > > +/**
> > > > + * Create a block device so U-Boot can access an EFI device
> > > > + *
> > > > + * @handle:        EFI handle to bind
> > > > + * @blkio: block io protocol
> > > > + * Return: 0 = success
> > > > + */
> > > > +int efi_bind_block(efi_handle_t handle, struct efi_block_io *blkio)
> > > > +{
> > > > +   struct efi_media_plat plat;
> > > > +   struct udevice *dev;
> > > > +   char name[18];
> > > > +   int ret;
> > > > +
> > > > +   plat.handle = handle;
> > > > +   plat.blkio = blkio;
> > > > +   ret = device_bind(dm_root(), DM_DRIVER_GET(efi_media), "efi_media",
> > > > +                     &plat, ofnode_null(), &dev);
> > > > +   if (ret)
> > > > +           return log_msg_ret("bind", ret);
> > > > +
> > > > +   snprintf(name, sizeof(name), "efi_media_%x", dev_seq(dev));
> > > > +   device_set_name(dev, name);
> > > > +
> > > > +   return 0;
> > > > +}
> > > > +
> > > >   static efi_status_t setup_memory(struct efi_priv *priv)
> > > >   {
> > > >     struct efi_boot_services *boot = priv->boot;
> > > > @@ -105,6 +135,95 @@ static void free_memory(struct efi_priv *priv)
> > > >     global_data_ptr = NULL;
> > > >   }
> > > >
> > > > +static int setup_disks(void)
> > > > +{
> > > > +   /* This is not fully implemented yet */
> > > > +   return 0;
> > > > +
> > > > +   efi_guid_t efi_disk_guid = EFI_DISK_IO_PROTOCOL_GUID;
> > > > +   struct efi_boot_services *boot = efi_get_boot();
> > > > +   struct efi_disk *disk;
> > > > +   int ret;
> > > > +
> > > > +   if (!boot)
> > > > +           return log_msg_ret("sys", -ENOSYS);
> > > > +   ret = boot->locate_protocol(&efi_disk_guid, NULL, (void **)&disk);
> > > > +   if (ret)
> > > > +           return log_msg_ret("prot", -ENOTSUPP);
> > > > +
> > > > +   return 0;
> > > > +}
> > > > +
> > > > +static int setup_block(void)
> > > > +{
> > > > +   efi_guid_t efi_blkio_guid = EFI_BLOCK_IO_PROTOCOL_GUID;
> > > > +   efi_guid_t efi_devpath_guid = EFI_DEVICE_PATH_PROTOCOL_GUID;
> > > > +   struct efi_boot_services *boot = efi_get_boot();
> > > > +   struct efi_block_io *blkio;
> > > > +   struct efi_device_path device_path;
> > > > +   efi_handle_t handle[100];
> > > > +   efi_uintn_t buf_size;
> > > > +   int num_handles;
> > > > +   int ret, i;
> > > > +
> > > > +   if (!boot)
> > > > +           return log_msg_ret("sys", -ENOSYS);
> > > > +
> > > > +   buf_size = sizeof(handle);
> > > > +   ret = boot->locate_handle(BY_PROTOCOL, &efi_blkio_guid, NULL,
> > > > +                             &buf_size, handle);
> > >
> > > You could use LocateHandleBuffer() here which will allocate enough
> > > memory for all matching handles.
> > >
> > > > +   if (ret)
> > > > +           return log_msg_ret("loc", -ENOTSUPP);
> > > > +
> > > > +   num_handles = buf_size / sizeof(efi_handle_t);
> > > > +   log_info("Found %d EFI handles\n", num_handles);
> > > > +
> > > > +   for (i = 0; i < num_handles; i++) {
> > > > +           ret = boot->handle_protocol(handle[i], &efi_devpath_guid,
> > > > +                                       (void **)&device_path);
> > >
> > >
> > > Why do you read the devicepath if you don't use it?
> > >
> > > > +           if (ret) {
> > > > +                   log_warning("- devpath %d failed (ret=%d)\n", i, ret);
> > > > +                   continue;
> > > > +           }
> > >
> > > Here some analysis of devicepaths and installed protocols is missing to
> > > find out which of the handles represents a block device and which
> > > represents a partition:
> > >
> > > If the last devicepath node is type 4 , Media Device Path with SubType 1
> > > Hard Drive and the partition number is non-zero it is a partition.
> > >
> > > If the devicepath without the last node relates to a handle with the
> > > EFI_BLOCK_IO_PROTOCOL, this also indicates that the current handle is
> > > for a partition.
> > >
> > > Best regards
> > >
> > > Heinrich
> > >
> > > > +
> > > > +           ret = boot->handle_protocol(handle[i], &efi_blkio_guid,
> > > > +                                       (void **)&blkio);
> > > > +           if (ret) {
> > > > +                   log_warning("- blkio %d failed (ret=%d)\n", i, ret);
> > > > +                   continue;
> > > > +           }
> > > > +
> > > > +           ret = efi_bind_block(handle[i], blkio);
> >
> > Here you are trying to create a U-Boot block device for
> > every UEFI block device (i.e.BLOCK_IO_PROTOCOL interface),
> > but please remember that U-Boot UEFI has created UEFI block devices
> > for all the existing U-Boot block devices at the initialization time.
> 
> Do you mean with EFI_LOADER? This code is for U-Boot as an app, where
> EFI_LOADER is disabled, at least at present.

Ah ok, but in this patch series, you touched some files under lib/efi_loader
and lib/efi_driver and I thought you have assumed the option as well.
So your patch#11 are fully independent from the rest.

-Takahiro Akashi

> >
> > So any physical disk may end up having one (original) U-Boot block device
> > and another U-Boot block device rooted in UEFI object (and
> > yet another U-Boot block device rooted in the second one and so on?).
> 
> Hopefully that is not an issue, for now. We will need to figure it out
> later, if we enable EFI_LOADER.
> 
> Regards,
> Simon
> 
> [..]

  reply	other threads:[~2021-09-10  0:51 UTC|newest]

Thread overview: 82+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-09-08 13:33 [PATCH 00/35] efi: Improvements to U-Boot running on top of UEFI Simon Glass
2021-09-08 13:33 ` [PATCH 01/35] x86: Keep symbol information in u-boot ELF file Simon Glass
2021-09-08 13:33 ` [PATCH 02/35] x86: Create a new header for EFI Simon Glass
2021-09-08 17:22   ` Heinrich Schuchardt
2021-09-09  8:57     ` Simon Glass
2021-09-09  9:25       ` Heinrich Schuchardt
2021-09-09 19:57         ` Simon Glass
2021-09-08 13:33 ` [PATCH 03/35] x86: Show some EFI info with the bdinfo command Simon Glass
2021-09-08 17:29   ` Heinrich Schuchardt
2021-09-09  8:57     ` Simon Glass
2021-09-09  9:29       ` Heinrich Schuchardt
2021-09-09 20:08         ` Simon Glass
2021-09-08 13:33 ` [PATCH 04/35] x86: Tidy up global_data pointer for 64-bit Simon Glass
2021-09-08 17:30   ` Heinrich Schuchardt
2021-09-08 13:33 ` [PATCH 05/35] efi: Add a script for building and testing U-Boot on UEFI Simon Glass
2021-09-08 13:33 ` [PATCH 06/35] x86: Create a 32/64-bit selection for the app Simon Glass
2021-09-08 17:35   ` Heinrich Schuchardt
2021-09-09  8:57     ` Simon Glass
2021-09-08 13:33 ` [PATCH 07/35] efi: Create a 64-bit app Simon Glass
2021-09-08 17:37   ` Heinrich Schuchardt
2021-09-08 13:33 ` [PATCH 08/35] x86: Don't duplicate global_ptr in 64-bit EFI app Simon Glass
2021-09-08 13:33 ` [PATCH 09/35] efi: Add a way to obtain boot services in the app Simon Glass
2021-09-08 13:33 ` [PATCH 10/35] efi: Add video support to " Simon Glass
2021-09-08 17:40   ` Heinrich Schuchardt
2021-09-09  8:57     ` Simon Glass
2021-09-08 13:33 ` [PATCH 11/35] RFC: efi: Drop code that doesn't work with driver model Simon Glass
2021-09-08 17:44   ` Heinrich Schuchardt
2021-09-09  8:57     ` Simon Glass
2021-09-09  9:21       ` Heinrich Schuchardt
2021-09-09 19:57         ` Simon Glass
2021-09-09 20:14           ` Tom Rini
2021-09-09 20:15           ` Mark Kettenis
2021-09-09 20:23             ` Tom Rini
2021-09-09 21:45               ` Mark Kettenis
2021-09-09 22:06                 ` Tom Rini
2021-09-24  2:48                 ` Simon Glass
2021-09-24 10:36                   ` Mark Kettenis
2021-09-24 12:32         ` Simon Glass
2021-09-08 13:33 ` [PATCH 12/35] efi: Add EFI uclass for media Simon Glass
2021-09-08 17:50   ` Heinrich Schuchardt
2021-09-09  8:57     ` Simon Glass
2021-09-08 13:33 ` [PATCH 13/35] efi: Add a media/block driver for EFI block devices Simon Glass
2021-09-08 17:59   ` Heinrich Schuchardt
2021-09-09  8:57     ` Simon Glass
2021-09-09 10:35       ` Heinrich Schuchardt
2021-09-09 19:58         ` Simon Glass
2021-09-08 13:33 ` [PATCH 14/35] efi: Locate all block devices in the app Simon Glass
2021-09-08 18:14   ` Heinrich Schuchardt
2021-09-09  1:11     ` AKASHI Takahiro
2021-09-09 20:09       ` Simon Glass
2021-09-10  0:50         ` AKASHI Takahiro [this message]
2021-09-09 20:09     ` Simon Glass
2021-09-08 13:33 ` [PATCH 15/35] patman: Use a ValueError exception if tools.Run() fails Simon Glass
2021-09-08 13:33 ` [PATCH 16/35] binman: Report an error if test files fail to compile Simon Glass
2021-09-08 13:33 ` [PATCH 17/35] binman: Support reading the offset of an ELF-file symbol Simon Glass
2021-09-08 13:33 ` [PATCH 18/35] binman: Allow timeout to occur in the image or its section Simon Glass
2021-09-08 13:33 ` [PATCH 19/35] binman: Tidy up comments on _DoTestFile() Simon Glass
2021-09-08 13:33 ` [PATCH 20/35] binman: Support updating the dtb in an ELF file Simon Glass
2021-09-08 13:33 ` [PATCH 21/35] efi: serial: Support arrow keys Simon Glass
2021-09-08 13:33 ` [PATCH 22/35] bloblist: Move to rST format Simon Glass
2021-09-08 18:18   ` Heinrich Schuchardt
2021-09-08 13:33 ` [PATCH 23/35] bloblist: Support allocating the bloblist Simon Glass
2021-09-08 13:33 ` [PATCH 24/35] x86: Allow booting a kernel from the EFI app Simon Glass
2021-09-08 18:22   ` Heinrich Schuchardt
2021-09-09  8:57     ` Simon Glass
2021-09-08 13:33 ` [PATCH 25/35] x86: Don't process the kernel command line unless enabled Simon Glass
2021-09-08 13:33 ` [PATCH 26/35] x86: efi: Add room for the binman definition in the dtb Simon Glass
2021-09-08 13:33 ` [PATCH 27/35] efi: Add comments to struct efi_priv Simon Glass
2021-09-08 13:33 ` [PATCH 28/35] efi: Fix ll_boot_init() operation with the app Simon Glass
2021-09-08 13:33 ` [PATCH 29/35] efi: Add a few comments to the stub Simon Glass
2021-09-08 13:34 ` [PATCH 30/35] efi: Share struct efi_priv between the app and stub code Simon Glass
2021-09-08 13:34 ` [PATCH 31/35] efi: Move exit_boot_services into a function Simon Glass
2021-09-08 13:34 ` [PATCH 32/35] efi: Check for failure when initing the app Simon Glass
2021-09-08 13:34 ` [PATCH 33/35] efi: Mention that efi_info_get() is only used in the stub Simon Glass
2021-09-08 13:34 ` [PATCH 34/35] efi: Show when allocated pages are used Simon Glass
2021-09-08 18:25   ` Heinrich Schuchardt
2021-09-09  8:57     ` Simon Glass
2021-09-09 10:39       ` Heinrich Schuchardt
2021-09-09 19:58         ` Simon Glass
2021-09-08 13:34 ` [PATCH 35/35] efi: Allow easy selection of serial-only operation Simon Glass
2021-09-09 16:29 ` [PATCH 00/35] efi: Improvements to U-Boot running on top of UEFI Bin Meng
2021-09-09 16:34   ` Tom Rini

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=20210910005046.GA42777@laputa \
    --to=takahiro.akashi@linaro.org \
    --cc=agraf@csgraf.de \
    --cc=bmeng.cn@gmail.com \
    --cc=christian.melki@t2data.com \
    --cc=ilias.apalodimas@linaro.org \
    --cc=sjg@chromium.org \
    --cc=trini@konsulko.com \
    --cc=u-boot@lists.denx.de \
    --cc=xypron.glpk@gmx.de \
    /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;
as well as URLs for NNTP newsgroup(s).