blk_get_device_by_str was skipping part_init when hw partition 0 was
selected because it is the default. However, this caused issues when
switching to a non-zero partition and then back to partition zero, as
stale data from the wrong partition was returned.
Remove this optimization and call part_init regardless of the selected
partition.
Signed-off-by: Robert Hancock <hancock@sedsystems.ca>
When compiling with FTRACE=1 an error
ld.bfd: disk/built-in.o: in function `lba512_muldiv':
disk/part.c:114: undefined reference to `__udivmoddi4
occurred.
Use '>> 11' instead of '/ 2048' to avoid the division.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Below is what happens on R-Car H3ULCB-KF using clean U-Boot
v2019.04-00810-g6aebc0d11a10 and r8a7795_ulcb_defconfig:
=> ### interrupt autoboot
=> gpt verify mmc 1
No partition list provided - only basic check
Verify GPT: success!
=> ### keep calling 'gpt verify mmc 1'
=> ### on 58th call, we are out of memory:
=> gpt verify mmc 1
alloc_read_gpt_entries: ERROR: Can't allocate 0X4000 bytes for GPT Entries
GPT: Failed to allocate memory for PTE
gpt_verify_headers: *** ERROR: Invalid Backup GPT ***
Verify GPT: error!
This is caused by calling is_gpt_valid() twice (hence allocating pte
also twice via alloc_read_gpt_entries()) while freeing pte only _once_
in the caller of gpt_verify_headers(). Fix that by freeing the pte
allocated and populated for primary GPT _before_ allocating and
populating the pte for backup GPT. The latter will be freed by the
caller of gpt_verify_headers().
With the fix applied, the reproduction scenario [1-2] has been run
hundreds of times in a loop w/o running into OOM.
[1] gpt verify mmc 1
[2] gpt verify mmc 1 $partitions
Fixes: cef68bf904 ("gpt: part: Definition and declaration of GPT verification functions")
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Below is what happens on R-Car H3ULCB-KF using clean U-Boot
v2019.04-00810-g6aebc0d11a10 and r8a7795_ulcb_defconfig:
=> ### interrupt autoboot
=> gpt guid mmc 1
21200400-0804-0146-9dcc-a8c51255994f
success!
=> ### keep calling 'gpt guid mmc 1'
=> ### on 59th call, we are out of memory:
=> gpt guid mmc 1
alloc_read_gpt_entries: ERROR: Can't allocate 0X4000 bytes for GPT Entries
GPT: Failed to allocate memory for PTE
get_disk_guid: *** ERROR: Invalid GPT ***
alloc_read_gpt_entries: ERROR: Can't allocate 0X4000 bytes for GPT Entries
GPT: Failed to allocate memory for PTE
get_disk_guid: *** ERROR: Invalid Backup GPT ***
error!
After some inspection, it looks like get_disk_guid(), added via v2017.09
commit 73d6d18b71 ("GPT: add accessor function for disk GUID"),
unlike other callers of is_gpt_valid(), doesn't free the memory pointed
out by 'gpt_entry *gpt_pte'. The latter is allocated by is_gpt_valid()
via alloc_read_gpt_entries().
With the fix applied, the reproduction scenario has been run hundreds
of times ('while true; do gpt guid mmc 1; done') w/o running into OOM.
Fixes: 73d6d18b71 ("GPT: add accessor function for disk GUID")
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
This adds a new block interface type for VirtIO block devices.
Signed-off-by: Tuomas Tynkkynen <tuomas.tynkkynen@iki.fi>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The current name conflicts with the Chrome OS verified boot library, which
prevents it being built. That library uses a string whereas U-Boot uses a
64-bit hex value. Rename this in U-Boot.
Signed-off-by: Simon Glass <sjg@chromium.org>
In int-ll64.h, we always use the following typedefs:
typedef unsigned int u32;
typedef unsigned long uintptr_t;
typedef unsigned long long u64;
This does not need to match to the compiler's <inttypes.h>.
Do not include it.
The use of PRI* makes the code super-ugly. You can simply use
"l" for printing uintptr_t, "ll" for u64, and no modifier for u32.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Underlying API should already print some meaningful error message, so
this one is just brings more noise. E.g. we can see log like this:
MMC: no card present
** Bad device mmc 0 **
Obviously, second error message is unwanted. Let's only print it in case
when DEBUG is defined to keep log short and clear.
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
When building with -pedantic the current definition of EFI_GUID() causes
an error 'initializer element is not constant'.
Currently EFI_GUID() is used both as an anonymous constant and as an
intializer. A conversion to efi_guid_t is not allowable when using
EFI_GUID() as an initializer. But it is needed when using it as an
anonymous constant.
We should not use EFI_GUID() for anything but an initializer. So let's
introduce a variable where needed and remove the conversion.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
In commit e163a931af ("cmd: gpt: backup boot code before writing MBR")
there was added the procedure for storing old boot code when doing "gpt
write". But instead of storing just backup code, the whole MBR was
stored, and only specific fields were replaced further, keeping
everything else intact. That's obviously not what we want.
Fix the code to actually store only old boot code and zero out
everything else. This fixes next testing case:
=> mmc write $loadaddr 0x0 0x7b
=> gpt write mmc 1 $partitions
In case when $loadaddr address and further memory contains 0xff, the
board was bricked (ROM-code probably didn't like partition entries that
were clobbered with 0xff). With this patch applied, commands above don't
brick the board.
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Cc: Alejandro Hernandez <ajhernandez@ti.com>
Tested-by: Andy Shevchenko <andy.shevchenko@gmail.com>
When U-Boot started using SPDX tags we were among the early adopters and
there weren't a lot of other examples to borrow from. So we picked the
area of the file that usually had a full license text and replaced it
with an appropriate SPDX-License-Identifier: entry. Since then, the
Linux Kernel has adopted SPDX tags and they place it as the very first
line in a file (except where shebangs are used, then it's second line)
and with slightly different comment styles than us.
In part due to community overlap, in part due to better tag visibility
and in part for other minor reasons, switch over to that style.
This commit changes all instances where we have a single declared
license in the tag as both the before and after are identical in tag
contents. There's also a few places where I found we did not have a tag
and have introduced one.
Signed-off-by: Tom Rini <trini@konsulko.com>
If SPL serial support is disabled nothing brings in sprintf, snprintf
or simple_strtoul:
env/built-in.o: In function `regex_callback':
env/attr.c:128: undefined reference to `sprintf'
disk/built-in.o: In function `blk_get_device_by_str':
disk/part.c:386: undefined reference to `simple_strtoul'
disk/part.c:395: undefined reference to `simple_strtoul'
disk/built-in.o: In function `blk_get_device_part_str':
disk/part.c:522: undefined reference to `simple_strtoul'
disk/built-in.o: In function `part_set_generic_name':
disk/part.c:704: undefined reference to `sprintf'
drivers/built-in.o: In function `init_peripheral_ep':
drivers/usb/musb-new/musb_gadget.c:1826: undefined reference to `sprintf'
drivers/built-in.o: In function `musb_core_init':
drivers/usb/musb-new/musb_core.c:1451: undefined reference to `snprintf'
Add those dependencies here.
Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com>
We have a large number of places where while we historically referenced
gd in the code we no longer do, as well as cases where the code added
that line "just in case" during development and never dropped it.
Signed-off-by: Tom Rini <trini@konsulko.com>
We enabled CONFIG_ISO_PARTITION by default for distro boot, so that U-Boot
could load distro images that usually get shipped as iso images. These images
usually come with a board agnostic boot environment.
However, there is very little point in having ISO support enabled (for anyone
really) in SPL, as the whole idea of SPL is to load U-Boot proper which again
is board specific. So the fact that we enable ISO support in U-Boot proper does
not mean at all that we want ISO support in U-Boot SPL.
Hence, let's remove the Kconfig dependency. Along the way, let's also clean up
all those default configs that disabled SPL ISO support.
Signed-off-by: Alexander Graf <agraf@suse.de>
Do not need to scan disk every time when we get part info
by name.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
[trini: Fix build in !CONFIG_HAVE_BLOCK_DEVICE case]
Signed-off-by: Tom Rini <trini@konsulko.com>
If a DUT do not have partition table, and we write one with 'gpt write'
cmd, we should able to list the partition with 'part list' cmd.
It's reasonable to scan the disk again if the initial part_type is
unknown in case we just write a new one into disk.
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
config_fallbacks.h has some logic that sets HAVE_BLOCK_DEVICE
based on a list of enabled options. Moving HAVE_BLOCK_DEVICE to
Kconfig allows us to drastically shrink the logic in
config_fallbacks.h
Signed-off-by: Adam Ford <aford173@gmail.com>
[trini: Rename HAVE_BLOCK_DEVICE to CONFIG_BLOCK_DEVICE]
Signed-off-by: Tom Rini <trini@konsulko.com>
config_fallback.h has some logic that checks a variety of options
and selects LIB_UUID if it hasn't already been selected. This
will all LIB_UUID in Kconfig and select this option for the list
of options to allow us to remove the logic from fallbacks
Signed-off-by: Adam Ford <aford173@gmail.com>
Commit ff98cb9051 ("part: extract MBR signature from partitions")
blindly switched allocated by ALLOC_CACHE_ALIGN_BUFFER buffer type from
"unsigned char" to "legacy_mbr" which caused allocation of size =
(typeof(legacy_mbr) * dev_desc->blksize) instead of just space enough
for "legacy_mbr" structure.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Peter Jones <pjones@redhat.com>
Cc: Alexander Graf <agraf@suse.de>
Cc: Tom Rini <trini@konsulko.com>
The SPL must be located at 8kB (16 sectors) offset. That's right in the
middle of the GPT, so we need to define a smaller amount of partitions to
accomodate for that location.
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
On some SoCs, the SPL needs to be located right in the middle of the GPT
partition entries.
One way to work around that is to create partition entries for a smaller
number of partitions to accomodate with where the SPL will be. Create a
Kconfig option to allow to do that.
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
use ALLOC_CACHE_ALIGN_BUFFER_PAD for mbr header allocation
in stack to fix alloc issue in is_gpt_valid()
this patch fix also issue for GPT partition handling
with blocksize != 512 in set_protective_mbr()
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
The check in part_get_info_extended() for a successful partition
searching misses a condition for extended partition. In case of
(ext_part_sector == 0), we should anyway mark the partition as found,
even if it's an extended partition, i.e. (is_extended(pt->sys_ind) == 0).
Otherwise, the extended partition (type 0x0f) will never be identified,
and the following recursive call to part_get_info_extended() will get a
wrong 'part_num' and 'which_part' parameter. In the end, all those
partitions in extended table will not be identified.
Let's add the missing OR condition of (ext_part_sector == 0) for
is_extended() check to fix the problem.
The issue is discovered by running fastboot flash to an extended
partition on eMMC.
$ fastboot flash mmcsda5 cache.img
target reported max download size of 536870912 bytes
sending 'mmcsda5' (18796 KB)...
OKAY [ 2.144s]
writing 'mmcsda5'...
FAILED (remote: cannot find partition)
finished. total time: 2.261s
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Before this patch one could receive following errors when executing
"gpt write" command on machine with cache enabled:
display5 factory > gpt write mmc ${mmcdev} ${partitions}
Writing GPT:
CACHE: Misaligned operation at range [4ef8f7f0, 4ef8f9f0]
CACHE: Misaligned operation at range [4ef8f9f8, 4ef939f8]
CACHE: Misaligned operation at range [4ef8f9f8, 4ef939f8]
CACHE: Misaligned operation at range [4ef8f7f0, 4ef8f9f0]
success!
To alleviate this problem - the calloc()s have been replaced with
malloc_cache_aligned() and memset().
After those changes the buffers are properly aligned (with both start
address and size) to SoC cache line.
Signed-off-by: Lukasz Majewski <lukma@denx.de>
the partition starting at 0x4400 is refused with overlap error:
$> gpt write mmc 0 "name=test,start=0x4400,size=0"
Writing GPT: Partition overlap
error!
even if the 0x4400 is the first available offset for LBA35 with default
value:
- MBR=LBA1
- GPT header=LBA2
- PTE= 32 LBAs (128 entry), 3 to 34
And the command to have one partition for all the disk failed also :
$> gpt write mmc 0 "name=test,size=0"
After the patch :
$> gpt write mmc 0 "name=test,size=0"
Writing GPT: success!
$> part list mmc 0
Partition Map for MMC device 0 -- Partition Type: EFI
Part Start LBA End LBA Name
Attributes
Type GUID
Partition GUID
1 0x00000022 0x01ce9fde "test"
attrs: 0x0000000000000000
type: ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
type: data
guid: b4b84b8a-04e3-4000-0036-aff5c9c495b1
And 0x22 = 34 LBA => offset = 0x4400 is accepted as expected
Reviewed-by: Łukasz Majewski <lukma@denx.de>
Tested-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
The test and info callbacks into the partition callback struct are
used by the "part list" command on the command line. That command
is used by the distro script.
With verb=1 set, "part list" thus throws a lot of warnings about
partitions it can't find when an upper layer searches for partitions.
So let's reduce verbosity to bring it to the same level of noise
as the other partition targets.
Signed-off-by: Alexander Graf <agraf@suse.de>
While this goes somewhat against normal coding style we should ensure
that dev_desc is not NULL before we dereference it in allocation of
legacy_mbr.
Reported-by: Coverity (CID: 167292)
Signed-off-by: Tom Rini <trini@konsulko.com>
There is already existing function part_get_info_by_name().
But sometimes user is particularly interested in looking for only
specific partition type. This patch implements such an API that
provides partition searching by name for specified partition type.
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Since commit ff98cb9051 ("part: extract MBR signature from partitions")
SPL boot on i.MX6 starts to fail:
U-Boot SPL 2017.09-00221-g0d6ab32 (Oct 02 2017 - 15:13:19)
Trying to boot from MMC1
(keep in loop)
Use the original allocation scheme for the SPL case, so that MX6 boards
can boot again.
This is a temporary solution to avoid the boot regression.
Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Acked-by: Rob Clark <robdclark@gmail.com>
Tested-by: Peter Robinson <pbrobinson@gmail.com>
U-Boot widely uses error() as a bit noisier variant of printf().
This macro causes name conflict with the following line in
include/linux/compiler-gcc.h:
# define __compiletime_error(message) __attribute__((error(message)))
This prevents us from using __compiletime_error(), and makes it
difficult to fully sync BUILD_BUG macros with Linux. (Notice
Linux's BUILD_BUG_ON_MSG is implemented by using compiletime_assert().)
Let's convert error() into now treewide-available pr_err().
Done with the help of Coccinelle, excluing tools/ directory.
The semantic patch I used is as follows:
// <smpl>
@@@@
-error
+pr_err
(...)
// </smpl>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
[trini: Re-run Coccinelle]
Signed-off-by: Tom Rini <trini@konsulko.com>
EFI client programs need the signature information from the partition
table to determine the disk a partition is on, so we need to fill that
in here.
Signed-off-by: Peter Jones <pjones@redhat.com>
[separated from efi_loader part, and fixed build-errors for non-
CONFIG_EFI_PARTITION case]
Signed-off-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
The UEFI spec allows an EFI system partition (ESP, with the bootloader or
kernel EFI apps on it) to reside on a disk using a "legacy" MBR
partitioning scheme.
But in contrast to actual legacy disks the ESP is not marked as
"bootable" using bit 7 in byte 0 of the legacy partition entry, but is
instead using partition *type* 0xef (in contrast to 0x0b or 0x0c for a
normal FAT partition). The EFI spec isn't 100% clear on this, but it even
seems to discourage the use of the bootable flag for ESPs.
Also it seems that some EFI implementations (EDK2?) even seem to ignore
partitions marked as bootable (probably since they believe they contain
legacy boot code).
The Debian installer [1] (*not* mini.iso), for instance, contains such an
MBR, where none of the two partitions are marked bootable, but the ESP
has clearly type 0xef.
Now U-Boot cannot find the ESP on such a disk (USB flash drive) and
fails to load the EFI grub and thus the installer.
Since it all boils down to the distro bootcmds eventually calling
"part list -bootable" to find potential boot partitions, it seems logical
to just add this "partition type is 0xef" condition to the is_bootable()
implementation.
This allows the bog standard arm64 Debian-testing installer to boot from
an USB pen drive on Allwinner A64 boards (Pine64, BananaPi-M64).
(Ubuntu and other distribution installers don't have a legacy MBR, so
U-Boot falls back to El Torito there).
[1] https://cdimage.debian.org/cdimage/daily-builds/daily/arch-latest/arm64/iso-cd/
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
Needed to support efi file protocol. The fallback.efi loader wants
to be able to read the contents of the /EFI directory to find an OS
to boot.
Modelled after POSIX opendir()/readdir()/closedir(). Unlike the other
fs APIs, this is stateful (ie. state is held in the FS_DIR "directory
stream"), to avoid re-traversing of the directory structure at each
step. The directory stream must be released with closedir() when it
is no longer needed.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Łukasz Majewski <lukma@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
The current code checks that no partitions overlap with the GPT partition
table using the offset of the first LBA usable for that partition.
This works fine, unless you have a partition entry that is further away
than it usually is and you want to create partitions in the gap between the
GPT header and the GPT partition entries, for example to reflash a
bootloader that needs to be set there.
Rework the test to something a bit smarter that checks whether a partition
would overlap with either the GPT header or the partition entries, no
matter where it is on the disk.
Partitions that do not have a start LBA specified will still start at the
first LBA usable set in the GPT header, to avoid weird behaviours.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
The gpt_fill_pte will need to access the device block size. Let's pass the
device descriptor as an argument.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
The start variable is only used inside a loop, and is never affected inside
it, so it's a purely local variable.
In the same way the partition size is accessed several times, so we can
store it in a variable.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Both the config option and the DT options specify the offset to set the GPT
at in bytes, yet the code treats those values as block numbers.
Fix that.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
When using ISO partitions with a DMA enabled block device driver
reading the ISO partition leads to unaligned DMA operations:
CACHE: Misaligned operation at range [bffb7da8, bffb85a8]
Align the buffer to make sure we pass a buffer which works for
DMA operations.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
We are now using an env_ prefix for environment functions. Rename these
two functions for consistency. Also add function comments in common.h.
Quite a few places use getenv() in a condition context, provoking a
warning from checkpatch. These are fixed up in this patch also.
Suggested-by: Wolfgang Denk <wd@denx.de>
Signed-off-by: Simon Glass <sjg@chromium.org>
This adds a new uclass id and block interface type for NVMe.
Signed-off-by: Zhikang Zhang <zhikang.zhang@nxp.com>
Signed-off-by: Wenbin Song <wenbin.song@nxp.com>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: Jon Nettleton <jon@solid-run.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
In order to read the GPT, modify the partition name strings, and then
write out a new GPT, the disk GUID is needed. While there is an
existing accessor for the partition UUIDs, there is none yet for the
disk GUID.
Changes since v6: none.
Signed-off-by: Alison Chaiken <alison@peloton-tech.com>
Move MAX_SEARCH_PARTITIONS to part.h so that functions in cmd
directory can find it. At the same time, increase the value to
64 since some operating systems use many, and the resources
consumed by a larger value are minimal.
Changes since v6: none.
Signed-off-by: Alison Chaiken <alison@peloton-tech.com>
Similar to what blk_get_device_part_str() does, this patch makes
part_get_info_by_name() return the partition number in case of a match.
This is useful when the partition number is needed and not just the
descriptor.
Signed-off-by: Alex Deymo <deymo@google.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
