OSDN -- 开发和下载开源软件
简体中文翻译状态 translation status for zh

Android-x86
Fork
Donation

  • R/O
  • HTTP
  • SSH
  • HTTPS
Fork

external-efivar: 提交列表

external/efivar


pie-x86
RSS
Rev. 时间 作者
836461e pie-x86 q-x86 r-x86 android-x86-9.0-r1 android-x86-9.0-r2 2019-03-06 03:14:47 Javier Martinez Canillas

dp-message: fix efidp_ipv4_addr fields assignment

The efidp_ipv4_addr structure has some 4-byte array fields to store IPv4
addresses and network mask. But the efidp_make_ipv4() function wrongly
casts these as a char * before dereferencing them to store a value.

Instead, cast it to a uint32_t * so the 32-bit value is correctly stored.

Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
4e04afc 2019-03-06 03:14:47 Javier Martinez Canillas

ucs2.h: fix logic that checks for UCS-2 string termination

Currently the loop to count the lenght of the UCS-2 string ends if either
of the two bytes are 0, but 0 is a valid value for UCS-2 character codes.

So only break the loop when 0 is the value for both UCS-2 char bytes.

Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
fdb8034 2019-03-06 03:14:47 Javier Martinez Canillas

ucs2.h: remove unused variable

The const uint16_t pointer is not used since now the two bytes of the
UCS-2 chars are checked to know if is the termination of the string.

Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
0dad6d7 2019-03-06 03:13:26 Chih-Wei Huang

Fix another error of -Werror=address-of-packed-member

Android 9 clang complains:

external/efivar/src/dp-message.c:367:24: error: taking address of packed member '' of class or structure 'efidp_infiniband' may result in an unaligned pointer value [-Werror,-Waddress-of-packed-member]
(efi_guid_t *)&dp->infiniband.ioc_guid);
^~~~~~~~~~~~~~~~~~~~~~~
external/efivar/src/dp.h:76:19: note: expanded from macro 'format_guid'
memmove(&_guid, guid, sizeof(_guid)); \
^~~~
1 error generated.

Since commit c3c553d the fifth parameter of format_guid() is treated as
a const void *. The casting is unnecessary.

Signed-off-by: Chih-Wei Huang <cwhuang@linux.org.tw>
22da79a 2019-03-06 03:13:26 Chih-Wei Huang

Remove an unused function

This gets rid of an error from Android 9 clang:

external/efivar/src/linux.c:256:1: error: unused function 'supports_iface' [-Werror,-Wunused-function]
supports_iface(struct dev_probe *probe, enum interface_type iftype)
^
1 error generated.

Signed-off-by: Chih-Wei Huang <cwhuang@linux.org.tw>
1a0ea5d 2019-03-06 03:13:26 Chih-Wei Huang

Android: inital porting of libefivar

The static library is linked by efibootmgr.

Signed-off-by: Chih-Wei Huang <cwhuang@linux.org.tw>
25d7360 2019-03-06 03:13:26 Chih-Wei Huang

Define strdupa if it is not defined

Android does not include strdupa in <string.h>. Define strdupa if it has
not already been defined.

Signed-off-by: Chih-Wei Huang <cwhuang@linux.org.tw>
e39fb5b 2019-02-22 05:27:16 Peter Jones

Get rid of the arrows in our debug messages.

They're not *that* useful, and the code is clever and problematic.

Resolves github issue #124

Signed-off-by: Peter Jones <pjones@redhat.com>
c3c553d 2019-02-22 05:27:09 Peter Jones

Fix all the places -Werror=address-of-packed-member catches.

This gets rid of all the places GCC 9's -Werror=address-of-packed-member
flags as problematic.

Fixes github issue #123

Signed-off-by: Peter Jones <pjones@redhat.com>
8f80379 2019-01-08 03:44:39 Peter Jones

linux-pci-root: remove an unused assignment

scan-build gripes about this, and it's pointless, so it can go.

Signed-off-by: Peter Jones <pjones@redhat.com>
b98ba89 2019-01-08 01:27:02 Peter Jones

dp.h: make format_guid() handle misaligned guid pointers safely.

GCC 9 adds -Werror=address-of-packed-member, which causes us to see the
build error reported at
https://bugzilla.opensuse.org/show_bug.cgi?id=1120862 .

That bug report shows us the following:

In file included from dp.c:26:
dp.h: In function 'format_vendor_helper':
dp.h:120:37: error: taking address of packed member of 'struct <anonymous>' may result in an unaligned pointer value [-Werror=address-of-packed-member]
120 | format_guid(buf, size, off, label, &dp->hw_vendor.vendor_guid);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
dp.h:74:25: note: in definition of macro 'format_guid'
74 | _rc = efi_guid_to_str(guid, &_guidstr); \
| ^~~~
cc1: all warnings being treated as errors

This patch makes format_guid() use a local variable as a bounce buffer
in the case that the guid we're passed is aligned as chaotic neutral.

Note that this only fixes this instance and there may be others that bz
didn't show because it exited too soon, and I don't have a gcc 9 build
in front of me right now.

Signed-off-by: Peter Jones <pjones@redhat.com>
cb0b1c3 2019-01-08 00:30:23 Peter Jones

util.h: add unlikely() and likely() macros

Signed-off-by: Peter Jones <pjones@redhat.com>
c1d6b10 2018-12-06 01:17:42 Peter Jones

Bump version to 37

- Minor coverity fixes
- Improve ACPI device path formatting
- Add support for SOC devices that use FDT as their PCI root node
- Make devices we can't parse the "device" sysfs link for use DEV_ABBREV_ONLY
- Handle SCSI port numbers better
- Don't require an EUI for NVMe
- Fix the accidental requirement on ACPI UID nodes existing
- Add support for EMMC devices
- Add support for PCI root nodes without a device link in sysfs
- Add support for partitioned MD devices
- Fix partition number detection when the number isn't provided
- Add support for ACPI Generic Container and Embedded Controller root nodes
- Add limited support for SAS/SATA port expanders

Signed-off-by: Peter Jones <pjones@redhat.com>
8a217a5 2018-12-06 01:16:53 Peter Jones

Workaround libabigail weirness.

As per usual, two runs of libabigail produce two different results :/

This removes the extra set of arguments it generated from
efidp_make_file() in efivar-36, because abicheck is complaining about
them going away, even though they're duplicates *and* other functions
have the same problem in the .abixml file.

I really wish this worked better.

Signed-off-by: Peter Jones <pjones@redhat.com>
6fe928b 2018-12-06 01:15:42 Peter Jones

Update abixml for newer libabigail

This is the xml files for the "36" tag, updated with libabigail-1.4-2.fc29.x86_64

Signed-off-by: Peter Jones <pjones@redhat.com>
3e687d8 2018-10-02 04:30:06 Peter Jones

sas: handle port expanders at all.

Signed-off-by: Peter Jones <pjones@redhat.com>
0d63973 2018-10-02 03:33:20 Peter Jones

pci-root: actually create the device path node.

Signed-off-by: Peter Jones <pjones@redhat.com>
7775b2f 2018-10-02 03:31:06 Peter Jones

Support decoding ACPI Generic Container and Embedded Controller dp nodes

Signed-off-by: Peter Jones <pjones@redhat.com>
ebc0b14 2018-10-02 03:27:04 Peter Jones

dp-acpi: fix HID vs HID_EX formatting bug.

Signed-off-by: Peter Jones <pjones@redhat.com>
cff3564 2018-09-20 03:26:23 Peter Jones

Fix partition number detection when it's not provided.

We need to actually get the partition number from the child device when
we're called without it.

Resolves: rhbz#1616305

Signed-off-by: Peter Jones <pjones@redhat.com>
576f55b 2018-09-18 05:52:57 Peter Jones

Handle partition name parsing and formatting for partitioned md.

Signed-off-by: Peter Jones <pjones@redhat.com>
e83002b 2018-09-18 05:13:24 Peter Jones

Deal with devices that don't have a ->device link in sysfs

Signed-off-by: Peter Jones <pjones@redhat.com>
e56cf8d 2018-09-18 05:12:25 Peter Jones

Fix dev->probes intialization test

Signed-off-by: Peter Jones <pjones@redhat.com>
da30e9f 2018-09-11 04:02:23 Peter Jones

Fix another buggy fake acpi pci root driver

In this case, the platform driver that creates the PCI(e) root device
doesn't fill in its driver link, so we can't look up what driver is in
use - but since it's the root, it *really* doesn't matter. And in
general, we only really care if it's the last node in our path, because
that'll be the controller for the boot device anyway.

Signed-off-by: Peter Jones <pjones@redhat.com>
0c44e3d 2018-08-04 02:11:11 Peter Jones

linux-emmc: update for internal API breakage.

Alek wrote his code before I made debug() imply LOG_DEBUG, so we need to
change that here as well.

Signed-off-by: Peter Jones <pjones@redhat.com>
cdb840b 2018-08-04 02:11:11 Peter Jones

Update abidw for newer tools.

Signed-off-by: Peter Jones <pjones@redhat.com>
1f385ff 2018-08-04 02:11:11 Peter Jones

abignore: work around an abidw bug

Some versions of abignore mistakenly treat brackets as section headers,
thus ignoring the rest of the stanza after them. It's fixed upstream,
but I don't have the newer version yet. Moving the braces to be after
the change-kind and other things works around the issue.

Signed-off-by: Peter Jones <pjones@redhat.com>
d6a526a 2018-08-03 22:47:36 Alek Du

emmc_parser: add emmc_parser

Signed-off-by: Alek Du <alek.du@intel.com>
50d7b97 2018-08-03 22:46:05 Bernhard M. Wiedemann

makeguids: initialize memory

so that we do not write uninitialized memory into guids.bin and names.bin
which made the resulting libefivar.so.1.36 unreproducible.
See https://reproducible-builds.org/ for why this matters.
d8d7e54 2018-07-17 04:40:26 Peter Jones

Don't require NVME to have an EUI

Resolves: rhbz#1593784

Signed-off-by: Peter Jones <pjones@redhat.com>
Show on old repository browser