[Cc: -jayalk@intworks.biz as it bounces]

Dear Chuansheng,


Am 29.03.22 um 01:58 schrieb Liu, Chuansheng:

quoted

-----Original Message-----
From: Paul Menzel
Sent: Monday, March 28, 2022 2:15 PM

quoted

Am 28.03.22 um 02:58 schrieb Liu, Chuansheng:

quoted

quoted

-----Original Message-----

quoted

quoted

Sent: Saturday, March 26, 2022 4:11 PM

quoted

quoted

Am 17.03.22 um 06:46 schrieb Chuansheng Liu:

quoted

Easily hit the below list corruption:
==
list_add corruption. prev->next should be next (ffffffffc0ceb090), but
was ffffec604507edc8. (prev=ffffec604507edc8).
WARNING: CPU: 65 PID: 3959 at lib/list_debug.c:26
__list_add_valid+0x53/0x80
CPU: 65 PID: 3959 Comm: fbdev Tainted: G     U
RIP: 0010:__list_add_valid+0x53/0x80
Call Trace:
    <TASK>
    fb_deferred_io_mkwrite+0xea/0x150
    do_page_mkwrite+0x57/0xc0
    do_wp_page+0x278/0x2f0
    __handle_mm_fault+0xdc2/0x1590
    handle_mm_fault+0xdd/0x2c0
    do_user_addr_fault+0x1d3/0x650
    exc_page_fault+0x77/0x180
    ? asm_exc_page_fault+0x8/0x30
    asm_exc_page_fault+0x1e/0x30
RIP: 0033:0x7fd98fc8fad1
==

Figure out the race happens when one process is adding &page->lru into
the pagelist tail in fb_deferred_io_mkwrite(), another process is
re-initializing the same &page->lru in fb_deferred_io_fault(), which is
not protected by the lock.

This fix is to init all the page lists one time during initialization,
it not only fixes the list corruption, but also avoids INIT_LIST_HEAD()
redundantly.

Fixes: 105a940416fc ("fbdev/defio: Early-out if page is already enlisted")
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Chuansheng Liu <redacted>
---
    drivers/video/fbdev/core/fb_defio.c | 9 ++++++++-
    1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/video/fbdev/core/fb_defio.c

b/drivers/video/fbdev/core/fb_defio.c

quoted

quoted

quoted

index 98b0f23bf5e2..eafb66ca4f28 100644

--- a/drivers/video/fbdev/core/fb_defio.c
+++ b/drivers/video/fbdev/core/fb_defio.c

@@ -59,7 +59,6 @@ static vm_fault_t fb_deferred_io_fault(struct vm_fault *vmf)
    		printk(KERN_ERR "no mapping available\n");

    	BUG_ON(!page->mapping);
-	INIT_LIST_HEAD(&page->lru);
    	page->index = vmf->pgoff;

    	vmf->page = page;

@@ -220,6 +219,8 @@ static void fb_deferred_io_work(struct work_struct *work)
    void fb_deferred_io_init(struct fb_info *info)
    {
    	struct fb_deferred_io *fbdefio = info->fbdefio;
+	struct page *page;
+	int i;

    	BUG_ON(!fbdefio);
    	mutex_init(&fbdefio->lock);

@@ -227,6 +228,12 @@ void fb_deferred_io_init(struct fb_info *info)
    	INIT_LIST_HEAD(&fbdefio->pagelist);
    	if (fbdefio->delay == 0) /* set a default of 1 s */
    		fbdefio->delay = HZ;
+
+	/* initialize all the page lists one time */
+	for (i = 0; i < info->fix.smem_len; i += PAGE_SIZE) {
+		page = fb_deferred_io_page(info, i);
+		INIT_LIST_HEAD(&page->lru);
+	}
    }
    EXPORT_SYMBOL_GPL(fb_deferred_io_init);

Applying your patch on top of current Linus’ master branch, tty0 is
unusable and looks frozen. Sometimes network card still works, sometimes
not.

I don't see how the patch would cause below BUG call stack, need some time to
debug. Just few comments:
1. Will the system work well without this patch?

Yes, the framebuffer works well without the patch.

quoted

2. When you are sure the patch causes the regression you saw, please get free

to submit one reverted patch, thanks : )

I think you for patch wasn’t submitted yet – at least not pulled by Linus.

The patch has been in drm-tip, could you have a try with the latest drm-tip to see if the
Framebuffer works well, in that case, we could revert it in drm-tip then.

With drm-tip (drm-tip: 2022y-03m-29d-13h-14m-35s UTC integration 
manifest) everything works fine. (I had to disable amdgpu driver, as it 
failed to build.) Is anyone able to explain that?


Kind regards,

Paul