Commit Graph

6390 Commits

Author SHA1 Message Date
Ben Hutchings 7e79e9a13a README.Debian: Change more URLs to use https: scheme 2016-04-25 00:52:47 +02:00
Ben Hutchings 35602a868b linux-source: Fix up module signing configuration in included kernel config files
Our signing certificate isn't included in the source tarball and would
be pointless to include in custom kernels.  Custom kernels also won't
have a separate signing stage.  So remove our settings for
CONFIG_MODULE_SIG_ALL, CONFIG_MODULE_SIG_KEY and
CONFIG_SYSTEM_TRUSTED_KEYS.  This should cause custom kernels based on
the included configs to follow the upstream default for signing, which
is to use a new key pair for each build.
2016-04-25 00:48:09 +02:00
Ben Hutchings 56d41759a0 Add module signature fix to the changelog 2016-04-23 20:52:24 +02:00
Ben Hutchings 5f6cfd0660 [x86] sound/soc/intel: Enable many more drivers (Closes: #822267) 2016-04-22 22:19:57 +02:00
Ben Hutchings 8a0b066e92 Note closure of #821442 2016-04-22 16:46:48 +02:00
Ben Hutchings 2d9dcd6be6 atl2: Disable unimplemented scatter/gather feature (CVE-2016-2117) 2016-04-22 08:10:30 +01:00
Ben Hutchings b20f5e22b0 [armel/marvell] dts: kirkwood: fix SD slot default configuration for OpenRD (Closes: #811351) 2016-04-22 08:06:49 +01:00
Ben Hutchings fe835b64b1 Update to 4.5.2
As this includes changes to header_ops and dentry_operations, bump
the ABI number to 2.
2016-04-22 08:05:48 +01:00
Aurelien Jarno c5cec59895 [mips*] Emulate unaligned LDXC1 and SDXC1 instructions. 2016-04-22 00:01:29 +02:00
Salvatore Bonaccorso 9c63adf133 [x86] USB: usbip: fix potential out-of-bounds write (CVE-2016-3955) 2016-04-19 16:30:27 +02:00
Ben Hutchings 690df616c2 fs: Consolidate softdep declarations in each module 2016-04-14 20:40:33 +01:00
Salvatore Bonaccorso 92f972094e [x86] xen: suppress hugetlbfs in PV guests (CVE-2016-3961) 2016-04-14 20:57:52 +02:00
Ben Hutchings 95034e0acb Prepare to release linux (4.5.1-1). 2016-04-14 12:30:56 +01:00
Ben Hutchings 315743de11 [armhf] usb-modules: Add modules required for BeagleBoard-X15 (Closes: #815848)
Thanks to Vagrant Cascadian
2016-04-14 00:07:03 +01:00
Ben Hutchings debb03a542 Note that #817816 was fixed in 4.5.1 2016-04-14 00:05:26 +01:00
Ben Hutchings a8c2d3c699 nbd: Create size change events for userspace (Closes: #812487) 2016-04-13 23:23:32 +01:00
Ben Hutchings e3395a2405 [armhf] watchdog: Enable S3C2410_WATCHDOG as module (Closes: #819377) 2016-04-13 23:12:44 +01:00
Ben Hutchings f782f13c81 [armhf] Add support for octa-core big.LITTLE systems including Exynos (Closes: #819379) 2016-04-13 23:11:02 +01:00
Ben Hutchings 1d7f287b7a [x86] ACPI / processor: Request native thermal interrupt handling via _OSC (Closes: #817016, #819336) 2016-04-13 23:06:17 +01:00
Ben Hutchings f2cd4c8e5a Re-enabling ZONE_DMA fixes the b44 and floppy drivers too (Closes: #819385, #820890) 2016-04-13 23:06:00 +01:00
Ben Hutchings a6c63dfcf4 aufs: Update support patches to aufs4.5-20160328 (Closes: #819748) 2016-04-13 22:21:26 +01:00
Ben Hutchings df965c4112 fs: Add MODULE_SOFTDEP declarations for hard-coded crypto drivers (Closes: #819725) 2016-04-13 22:17:00 +01:00
Ben Hutchings aac56d9572 [x86] mm/32: Enable full randomization on i386 and X86_32 (CVE-2016-3672) 2016-04-13 21:26:06 +01:00
Ben Hutchings e01d7b854c ipv4: Don't do expensive useless work during inetdev destroy (CVE-2016-3156) 2016-04-13 21:24:19 +01:00
Ben Hutchings 3f1372dedd Add more CVE IDs for 4.5.1 2016-04-13 21:22:24 +01:00
Ben Hutchings d0292c6f67 netfilter: x_tables: Fix parsing of IPT_SO_SET_REPLACE blobs (CVE-2016-3134) 2016-04-13 21:11:40 +01:00
Ben Hutchings c972e924a7 Prepare for unstable upload: set ABI to 1 and drop ~exp1 from the version 2016-04-13 20:53:11 +01:00
Ben Hutchings 2b5668bad8 linux-support: Include udeb configuration from debian/installer
This will be used by the linux-signed package.
2016-04-13 20:53:05 +01:00
Ben Hutchings 74940e0a05 Disable UNUSED_SYMBOLS; it is now itself unused 2016-04-13 20:52:59 +01:00
Ben Hutchings 68c5c5997c Merge remote-tracking branch 'alioth/sid'
Merge open changelog entry for 4.4.6-2 into our open changelog entry.

Refresh the new patches.
2016-04-13 20:52:01 +01:00
Ben Hutchings 6e05e68d7a Update to 4.5.1 2016-04-13 20:49:29 +01:00
Roger Shimizu dfd9a83664 [armhf/armmp] Enable DW_WATCHDOG as module
Enable DW_WATCHDOG as module, used on Firefly-RK3288.
Patch by Vagrant Cascadian.
(Closes: #820834)
2016-04-13 22:37:43 +09:00
Roger Shimizu 5c3489a433 [armel/marvell] Add DT support for "Buffalo/Revogear Kurobox Pro" 2016-04-08 08:33:27 +09:00
Aurelien Jarno 062efc4b3f [mips*/octeon] Enable CRYPTO_MD5_OCTEON, CRYPTO_SHA1_OCTEON, CRYPTO_SHA256 and CRYPTO_SHA512_OCTEON. 2016-04-06 13:08:06 +02:00
Aurelien Jarno 4f0dd22673 [mips*/octeon] Bump CONFIG_NR_CPUS to 32 to accomodate recent Octeon III SoCs. 2016-04-06 12:27:15 +02:00
Aurelien Jarno c068d48b25 [mips*/octeon] udeb: Add ahci_octeon and ahci_platform modules to sata-modules. 2016-04-06 00:06:31 +02:00
Aurelien Jarno 511d4cedbc [mips*/octeon] Enable SPI_OCTEON. 2016-04-05 23:58:00 +02:00
Ben Hutchings 54ab3395af Group all securelevel and Secure Boot changes together, closing #820008 2016-04-05 14:03:57 +01:00
Ben Hutchings 119c44d06d Merge branch 'benh/secure-boot' 2016-04-05 13:59:42 +01:00
Aurelien Jarno 7836b549be [mips*/octeon] Backport OCTEON SATA controller support from 4.6-rc1. Enable AHCI_OCTEON. 2016-04-04 22:51:11 +02:00
Aurelien Jarno d514efc4cc [mipsel/mips/config.loongson-2f] Disable VIDEO_CX23885, VIDEO_IVTV, VIDEO_CX231XX, VIDEO_PVRUSB2 (fixes FTBFS).
This workaround the failed compilation of cx25840-core.c:

| {standard input}: Assembler messages:
| {standard input}:583: Error: branch out of range
| {standard input}:636: Error: branch out of range
| {standard input}:40094: Error: branch out of range

These are caused by the -Wa,-mfix-loongson2f-jump, which replaces the
"jalr t9" instruction by the "or t9, t9, 0x80000000; jalr t9"
instructions, thus causing an increase in code size.

This is probably the best way to workaround the FTBFS, especially given
that the loongson-2e/2f flavour are likely going to be removed for
stretch.
2016-04-04 22:20:30 +02:00
Ben Hutchings 76de9f06e0 scripts: Fix X.509 PEM support in sign-file
DER format works but it's easier if we can use PEM everywhere.
2016-04-04 19:28:26 +01:00
Ben Hutchings 6e18d075e9 linux-kbuild: Add extract-cert and sign-file programs 2016-04-04 10:41:48 +01:00
Ben Hutchings 23d1e0f3c2 debian/copyright: Note that extract-cert and sign-file are under LGPL 2.1
This is significant because they link with OpenSSL.
2016-04-04 10:41:48 +01:00
Ben Hutchings f880a7ff25 Fix config for module signing
Replace my pubkey with an X.509 (PEM encoded) certificate as actually
required.  Add quotes around the filenames in kconfig.
2016-04-04 10:41:46 +01:00
Ben Hutchings a6aaaeb263 Note added build-dependency on openssl too 2016-04-04 02:04:53 +01:00
Ben Hutchings 969431b952 certs: Set SYSTEM_TRUSTED_KEYS to my own personal key to support initial testing of signed modules
Also set MODULE_SIG_KEY to empty string to avoid including a build-
time generated key.
2016-04-03 13:12:23 +01:00
Ben Hutchings c955e35c32 modules: Enable MODULE_SIG and MODULE_SIG_SHA256
...but not MODULE_SIG_ALL as signatures will be packaged separately
2016-04-03 13:12:23 +01:00
Ben Hutchings 7321950826 Add Matthew Garrett's securelevel patchset in preparation for Secure Boot support 2016-04-03 04:31:52 +01:00
Ben Hutchings ea7af96b4d Merge branch 'benh/merge-linux-tools'
The primary reason not to build userland tools from src:linux was that
it allowed building with a minimal cross-toolchain and without the
target's glibc development files.  In particular, the CI system at
kernel-archive.buildserver.net used such minimal cross-toolchains.

It is been a long time since kernel-archive.buildserver.net was
running, so the original reason for the separation no longer exists.
If we implement CI using similarly limited toolchains again, we can use
a build-profile to exclude userland builds, as these are now well
supported.

This merges the full history of both source packages together,
including in debian/changelog (which looks rather weird).  The binary
packages resulting from this merge appear to be functionally the same,
though I can't be certain there's no regression.
2016-04-02 15:13:23 +01:00