4 coreboot 4.12 was released on May 12th, 2020.
6 Since 4.11 there were 2692 new commits by over 190 developers and of
7 these, 59 contributed for the first time, which is quite an amazing
10 Thank you to all developers who again helped made coreboot better
11 than ever, and a big welcome to our new contributors!
16 This release saw some activity on the MAINTAINERS file, showing more
17 persons, teams and companies declare publicly that they intend to
18 take care of mainboards and subsystems.
20 To all new maintainers, thanks a lot!
25 Our documentation efforts in the code tree are picking up steam, with
26 some 70 commits in that general area. Everything from typo fixes to
27 documenting mainboard support or coreboot APIs.
29 There's still room to improve, but the contributions are getting more
35 The removals due to the announced deprecations as well as the
36 deduplication of boards into variants skew the stats a bit, so at
37 a top level view this is a rare coreboot release in that it removes
38 more boards (51) than it adds (49).
40 After accounting for the variant moves the numbers in favor of more
41 hardware supported than the previous version. Besides a whole lot
42 of Chrome OS devices (again), this release features a whole bunch
43 of retrofits for devices originally shipping with non-coreboot OEM
44 firmware, but also support for devices that come with coreboot right
47 For that, a shout out to System76, Protectli, Libretrend and the
53 We simplified the header that comes at the top of every file:
54 Instead of a lengthy reference to the license any given file
55 is under, or even the license text itself, we opted for simple
56 [SPDX](https://www.spdx.org) identifiers.
58 Since people also handled copyright lines differently, we now opt for
59 collecting authors in AUTHORS and let git history tell the whole story.
61 While at it, the content-free "This file is part of this-and-that
62 project" header was also dropped.
64 Besides that, there has also been more work to sort out the headers
65 we include across the tree to minimize the code impacting every
68 Now that our board-variant mechanism matured, many boards that were
69 individual models so far were converted into variants, making it
70 easier to maintain families of devices.
75 For the 4.12 release a few features on x86 became mandatory. These are
76 relocatable ramstage, postcar stage and C\_ENVIRONMENT\_BOOTBLOCK.
78 ### Relocatable ramstage
80 Relocatable stages are a feature implemented only on x86, where stages
81 can be relocated at runtime. This is used to place ramstage in a better
82 location that does not collide with memory the OS or the payload tends
83 to use. The rationale behind making this mandatory is that you always
84 want cbmem to be cached so it's a good location to run ramstage from.
85 It avoids using lower memory altogether so the OS can make use of it
86 and no backing up needs to happen on S3 resume.
90 With Postcar stage tearing down Cache-as-Ram is done in a separate
91 stage. This means that romstage has a clean program boundary and
92 that all variables in romstage can be accessed via their linked
93 addresses without runtime resolution. There is no need to link
94 global and static variables via the CAR\_GLOBAL macro and no need
95 to access them with car\_set/get\_var/ptr functions.
97 ### C\_ENVIRONMENT\_BOOTBLOCK
99 Historically the bootblock on x86 platforms has been compiled with
100 romcc. This means that the generated code only uses CPU registers
101 and therefore no stack. This 20K+ LOC compiler is limited and hard
102 to maintain and so is the code that one has to write in that
103 environment. A different solution is to set up Cache-as-Ram in the
104 bootblock and run GCC compiled code in the bootblock. The advantages
105 are increased flexibility and consistency with other architectures as
106 well as other stages: e.g. printing to console is possible and
107 VBOOT can run before romstage, making romstage updatable via RW FMAP
110 ### Platforms dropped from master
112 The following platforms did not implement those feature are dropped
113 from master to allow the master branch to move on:
115 - all FSP1.0 platforms: BROADWELL\_DE, FSP\_BAYTRAIL, RANGELEY
118 In particular on FSP1.0 it is impossible to implement POSTCAR stage.
119 The reason is that FSP1.0 relocates the CAR region to the HOB before
120 returning to coreboot. This means that after FSP returns to coreboot
121 accessing variables via their original address is not possible. One
122 way of obtaining that behavior would be to set up Cache-as-Ram again
123 (but with open source code) and copy the relocated data from the HOB
124 there. This solution is deemed too hacky. Maybe a lesson can be
125 learned from this: blobs should not interfere with the execution
126 environment, as this makes proper integration much harder.
130 Given that some platforms supported by FSP1.0 are being produced and
131 popular, the 4.11 release was made into a branch in which further
132 development can happen.
137 ### SMMSTORE is now production ready
139 See [smmstore](../drivers/smmstore.md) for the documentation on
140 the API, but note that there will be an update to it featuring a
141 much-improved but incompatible API.
143 ### Unit testing infrastructure
145 Unit testing of coreboot is now possible in a more structured way, with new
146 build subsystem and adoption of [Cmocka](https://cmocka.org/) framework. Tree
147 has new directory `tests/`, which comprises infrastructure and examples of unit
149 [Unit testing coreboot](../technotes/2020-03-unit-testing-coreboot.md) for the
155 Your favorite new feature or supported board didn't make it to the
156 release notes? They're maintained collaboratively in the coreboot
157 tree, so when you land something noteworthy don't be shy, contribute
158 to the upcoming release's document in Documentation/releases!