+139
−187
Loading
Gitlab 现已全面支持 git over ssh 与 git over https。通过 HTTPS 访问请配置带有 read_repository / write_repository 权限的 Personal access token。通过 SSH 端口访问请使用 22 端口或 13389 端口。如果使用CAS注册了账户但不知道密码,可以自行至设置中更改;如有其他问题,请发邮件至 service@cra.moe 寻求协助。
The Freescale DIU driver allocates multiple blocks of memory, including multiple DMA buffers. Merge all of these blocks into one data structure. Specifically: 1) struct fsl_diu_data now contains everything that needs to be allocated, except for the framebuffers themselves. DMA'able objects are aligned correctly within the structure. 2) struct diu_addr is no longer needed, because we don't have to manage multiple blocks of DMA memory. 3) Since there's no diu_addr any more, macro DMA_ADDR is used to calculate the DMA address of any field in fsl_diu_data. 4) Functions allocate_buf() and free_buf() are no longer needed, because we now assume that dma_alloc_coherent() will allocate a page-aligned block, and everything is properly aligned with fsl_diu_data already, so we no longer need to align any memory blocks ourselves. 5) The "dummy" area descriptor is now defined separately from the other five ADs, so NUM_AOIS (previously called FSL_AOI_NUM) is now set to five instead of six. Previously, all six were combined together to avoid a separate call to allocate_buf() just for the dummy AD. 6) framebuffer_alloc() and framebuffer_release() are no longer used. The framebuffer is initialized manually. 7) Error handling is simplified since there's only one memory buffer allocated. Signed-off-by:Timur Tabi <timur@freescale.com> Signed-off-by:
Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
CRA Git | Maintained and supported by SUSTech CRA and CCSE