<!DOCTYPE html> <html> <head> <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> </head> <body> <div class="moz-cite-prefix">在 2025/6/27 16:12, Thomas Zimmermann 写道: </div> <blockquote type="cite" cite="mid:855d6faa-9f72-466e-9294-d6059bb9d920@suse.de">Hi Am 27.06.25 um 10:07 schrieb Shixiong Ou: <blockquote type="cite"> 在 2025/6/26 18:31, Thomas Zimmermann 写道: <blockquote type="cite">Hi Am 26.06.25 um 11:49 schrieb <a class="moz-txt-link-abbreviated" href="mailto:oushixiong1025@163.com">oushixiong1025@163.com</a>: <blockquote type="cite">From: Shixiong Ou <a class="moz-txt-link-rfc2396E" href="mailto:oushixiong@kylinos.cn"><oushixiong@kylinos.cn></a> [WHY] On an ARM machine, the following log is present: [ 0.900884] efifb: framebuffer at 0x1020000000, using 3072k, total 3072k [ 2.297884] amdgpu 0000:04:00.0: remove_conflicting_pci_framebuffers: bar 0: 0x1000000000 -> 0x100fffffff [ 2.297886] amdgpu 0000:04:00.0: remove_conflicting_pci_framebuffers: bar 2: 0x1010000000 -> 0x10101fffff [ 2.297888] amdgpu 0000:04:00.0: remove_conflicting_pci_framebuffers: bar 5: 0x58200000 -> 0x5823ffff It show that the efifb framebuffer base is out of PCI BAR, and this </blockquote> The patch at <a class="moz-txt-link-freetext" href="https://patchwork.freedesktop.org/series/148057/">https://patchwork.freedesktop.org/series/148057/</a> is supposed to fix the problem. It has been merged with v6.16-rc1 as commit 2f29b5c23101 ("video: screen_info: Relocate framebuffers behind PCI bridges"). It is in your tree? Best regards Thomas </blockquote> yeah, this patch is in my tree. but do not fix the problem. </blockquote> The patch's final revision had a rough development. Just for testing purposes, could you revert the commit and instead apply the patch's earlier revision from <a class="moz-txt-link-freetext" href="https://patchwork.freedesktop.org/patch/649527/?series=148057&rev=1">https://patchwork.freedesktop.org/patch/649527/?series=148057&rev=1</a> ? Thanks a lot. Best regards Thomas </blockquote> I have revert the commit and applied this patch, and added some prints as: + printk("pcibios_bus_to_resource orginal: start = %llx, end = %llx", + r->start, r->end); + pcibios_bus_to_resource(pdev->bus, r, &bus_region); + printk("pcibios_bus_to_resource finished: start = %llx, end = %llx", + r->start, r->end); + and the kernel message as follow: kylin@kylin-pc:~$ dmesg | grep pcibios_bus_to_resource [ 0.684698] pcibios_bus_to_resource orginal: start = 1020000000, end = 10202fffff [ 0.684702] pcibios_bus_to_resource finished: start = 1020000000, end = 10202fffff <div id="tta_output_ta" class="tta_no_click_outline tta_output_hastxt tta_focusTextExtraExtraSmall" dir="ltr" data-placeholder="翻译" style="height: 48px;" lang="zh-Hans" tabindex="0">The address doesn't seem to have been modified.</div> <div class="tta_no_click_outline tta_output_hastxt tta_focusTextExtraExtraSmall" dir="ltr" data-placeholder="翻译" style="height: 48px;" lang="zh-Hans" tabindex="0">Best regards Shixiong. </div> <blockquote type="cite" cite="mid:855d6faa-9f72-466e-9294-d6059bb9d920@suse.de"> <blockquote type="cite"> this is some message: kylin@kylin-pc:~$ dmesg | grep BAR [ 0.688192] pci 0000:00:03.0: BAR 15: assigned [mem 0x1000000000-0x101fffffff 64bit pref] [ 0.688200] pci 0000:00:00.0: BAR 0: assigned [mem 0x1020000000-0x10200fffff 64bit pref] [ 0.688205] pci 0000:00:00.0: BAR 14: assigned [mem 0x58000000-0x580fffff] [ 0.688210] pci 0000:00:01.0: BAR 0: assigned [mem 0x1020100000-0x10201fffff 64bit pref] [ 0.688215] pci 0000:00:02.0: BAR 0: assigned [mem 0x1020200000-0x10202fffff 64bit pref] [ 0.688221] pci 0000:00:02.0: BAR 14: assigned [mem 0x58100000-0x581fffff] [ 0.688225] pci 0000:00:03.0: BAR 0: assigned [mem 0x1020300000-0x10203fffff 64bit pref] [ 0.688231] pci 0000:00:03.0: BAR 14: assigned [mem 0x58200000-0x585fffff] [ 0.688237] pci 0000:00:04.0: BAR 0: assigned [mem 0x1020400000-0x10204fffff 64bit pref] [ 0.688243] pci 0000:00:05.0: BAR 0: assigned [mem 0x1020500000-0x10205fffff 64bit pref] [ 0.688249] pci 0000:00:05.0: BAR 14: assigned [mem 0x58600000-0x586fffff] [ 0.688253] pci 0000:01:00.0: BAR 0: assigned [mem 0x58000000-0x58003fff 64bit] [ 0.688290] pci 0000:03:00.0: BAR 6: assigned [mem 0x58100000-0x5817ffff pref] [ 0.688296] pci 0000:03:00.0: BAR 0: assigned [mem 0x58180000-0x58181fff] [ 0.688303] pci 0000:03:00.0: BAR 5: assigned [mem 0x58182000-0x58183fff] [ 0.688317] pci 0000:04:00.0: BAR 1: assigned [mem 0x1000000000-0x101fffffff 64bit pref] [ 0.688326] pci 0000:04:00.0: BAR 0: assigned [mem 0x58200000-0x583fffff] [ 0.688332] pci 0000:04:00.0: BAR 6: assigned [mem 0x58400000-0x584fffff pref] [ 0.688336] pci 0000:04:00.1: BAR 0: assigned [mem 0x58500000-0x58503fff] [ 0.688360] pci 0000:06:00.0: BAR 0: assigned [mem 0x58600000-0x58601fff 64bit] kylin@kylin-pc:~$ dmesg | grep framebuffer [ 1.137536] efifb: framebuffer at 0x1020000000, using 3072k, total 3072k the efifb base address is still at 0x1020000000 after calling pcibios_bus_to_resource(). <blockquote type="cite"> <blockquote type="cite">results in both efi-framebuffer and amdgpudrmfb co-existing. The fbcon will be bound to efi-framebuffer by default and cannot be used. [HOW] Do not load efifb driver if PCI BAR has changed but not fixuped. In the following cases: 1. screen_info_lfb_pdev is NULL. 2. __screen_info_relocation_is_valid return false. Signed-off-by: Shixiong Ou <a class="moz-txt-link-rfc2396E" href="mailto:oushixiong@kylinos.cn"><oushixiong@kylinos.cn></a> --- drivers/video/fbdev/efifb.c | 4 ++++ drivers/video/screen_info_pci.c | 24 ++++++++++++++++++++++++ include/linux/screen_info.h | 5 +++++ 3 files changed, 33 insertions(+) diff --git a/drivers/video/fbdev/efifb.c b/drivers/video/fbdev/efifb.c index 0e1bd3dba255..de8d016c9a66 100644 --- a/drivers/video/fbdev/efifb.c +++ b/drivers/video/fbdev/efifb.c @@ -303,6 +303,10 @@ static void efifb_setup(struct screen_info *si, char *options) static inline bool fb_base_is_valid(struct screen_info *si) { + /* check whether fb_base has changed but not fixuped */ + if (!screen_info_is_useful()) + return false; + if (si->lfb_base) return true; diff --git a/drivers/video/screen_info_pci.c b/drivers/video/screen_info_pci.c index 66bfc1d0a6dc..ac57dcaf0cac 100644 --- a/drivers/video/screen_info_pci.c +++ b/drivers/video/screen_info_pci.c @@ -9,6 +9,8 @@ static struct pci_dev *screen_info_lfb_pdev; static size_t screen_info_lfb_bar; static resource_size_t screen_info_lfb_res_start; // original start of resource static resource_size_t screen_info_lfb_offset; // framebuffer offset within resource +static bool screen_info_changed; +static bool screen_info_fixuped; static bool __screen_info_relocation_is_valid(const struct screen_info *si, struct resource *pr) { @@ -24,6 +26,24 @@ static bool __screen_info_relocation_is_valid(const struct screen_info *si, stru return true; } +bool screen_info_is_useful(void) +{ + unsigned int type; + const struct screen_info *si = &screen_info; + + type = screen_info_video_type(si); + if (type != VIDEO_TYPE_EFI) + return true; + + if (screen_info_changed && !screen_info_fixuped) { + pr_warn("The screen_info has changed but not fixuped"); + return false; + } + + pr_info("The screen_info is useful"); + return true; +} + void screen_info_apply_fixups(void) { struct screen_info *si = &screen_info; @@ -32,18 +52,22 @@ void screen_info_apply_fixups(void) struct resource *pr = &screen_info_lfb_pdev->resource[screen_info_lfb_bar]; if (pr->start != screen_info_lfb_res_start) { + screen_info_changed = true; if (__screen_info_relocation_is_valid(si, pr)) { /* * Only update base if we have an actual * relocation to a valid I/O range. */ __screen_info_set_lfb_base(si, pr->start + screen_info_lfb_offset); + screen_info_fixuped = true; pr_info("Relocating firmware framebuffer to offset %pa[d] within %pr\n", &screen_info_lfb_offset, pr); } else { pr_warn("Invalid relocating, disabling firmware framebuffer\n"); </blockquote> </blockquote> And should something be done after __screen_info_relocation_is_valid() return false? Best regards Shixiong. <blockquote type="cite"> <blockquote type="cite"> } } + } else { + screen_info_changed = true; } } diff --git a/include/linux/screen_info.h b/include/linux/screen_info.h index 923d68e07679..632cdbb1adbe 100644 --- a/include/linux/screen_info.h +++ b/include/linux/screen_info.h @@ -138,9 +138,14 @@ ssize_t screen_info_resources(const struct screen_info *si, struct resource *r, u32 __screen_info_lfb_bits_per_pixel(const struct screen_info *si); #if defined(CONFIG_PCI) +bool screen_info_is_useful(void); void screen_info_apply_fixups(void); struct pci_dev *screen_info_pci_dev(const struct screen_info *si); #else +bool screen_info_is_useful(void) +{ + return true; +} static inline void screen_info_apply_fixups(void) { } static inline struct pci_dev *screen_info_pci_dev(const struct screen_info *si) </blockquote> </blockquote> </blockquote> </blockquote> </body> </html>