Re: update firmware binary for v5.4
Renjiang Han <[email protected]>
| Newsgroups | org.kernel.lore.linux-firmware |
|---|---|
| Message-ID | <[email protected]> |
On 11/21/2025 7:55 PM, Dmitry Baryshkov wrote: > On Fri, Nov 21, 2025 at 03:17:24PM +0530, Vikash Garodia wrote: >> On 11/21/2025 9:27 AM, Renjiang Han wrote: >>> Hi Dmitry & Vikash >>> >>> I have re-submitted two changes, please help check. Thanks! >>> >>> The following changes since commit >>> 6fc940781a013ad837ed8fea326d2b897467bbc3: >>> >>> Merge branch 'robot/patch-0-1762826844' into 'main' (2025-11-11 >>> 12:02:02 +0000) >>> >>> are available in the Git repository at: >>> >>> https://git.codelinaro.org/clo/linux-kernel/linux-firmware.git update- >>> video-fw-v5.4 >>> >>> for you to fetch changes up to bb28dcbcf9867ebeda6a41d575656a737c24f32f: >>> >>> qcom: venus-5.4: update firmware binary for v5.4 (2025-11-21 08:55:41 >>> +0530) >>> >>> ---------------------------------------------------------------- >>> Renjiang Han (2): >>> qcom: venus-5.4: remove unused firmware file >>> qcom: venus-5.4: update firmware binary for v5.4 >>> >>> WHENCE | 6 +----- >>> qcom/venus-5.4/venus.mbn | Bin 922216 -> 922312 bytes >>> qcom/venus-5.4/venus_s2.mbn | Bin 919008 -> 0 bytes >>> 3 files changed, 1 insertion(+), 5 deletions(-) >>> delete mode 100644 qcom/venus-5.4/venus_s2.mbn >> It was made quite clear to put up the history from versions #58 all the way >> to #62 in the commit message, still i see commit message does not capture >> it. > No, it's not required to capture old history. But it would be nice to > point out the reasons for the change (conformance to ELF ABI). BTW: what > is the actual alignment requirement for those segments? E.g. on ARM > systems code segment must be alisnged at least to 0x4. Likewise data > segments also should have some alignemnt (again, 0x4 at least, usually > 0x8 or ox10), otherwise loading data from those segments becomes very > unefficient. Correct,the data segment needs to have proper alignment. However, the alignment value must comply with the ELF specification, which requires that (Offset % Alignment) equals (Virtual Address % Alignment). After we discussion and verification, we decided to use 0x4 alignment. So I will update new binary with 0x4 alignment. Thank you very much for your comment. -- Best regards, Renjiang Han