[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <Y6XZnz0EDXYlfqhX@infradead.org>
Date: Fri, 23 Dec 2022 08:38:55 -0800
From: Christoph Hellwig <hch@...radead.org>
To: Keith Busch <kbusch@...a.com>
Cc: linux-mm@...ck.org, linux-kernel@...r.kernel.org,
Matthew Wilcox <willy@...radead.org>,
Tony Battersby <tonyb@...ernetics.com>,
Kernel Team <kernel-team@...a.com>,
Keith Busch <kbusch@...nel.org>
Subject: Re: [PATCHv2 09/11] dmapool: simplify freeing
> @@ -280,14 +268,14 @@ void dma_pool_destroy(struct dma_pool *pool)
> mutex_unlock(&pools_reg_lock);
>
> list_for_each_entry_safe(page, tmp, &pool->page_list, page_list) {
> + if (!is_page_busy(page))
> + dma_free_coherent(pool->dev, pool->allocation,
> + page->vaddr, page->dma);
> + else
> dev_err(pool->dev, "%s %s, %p busy\n", __func__,
> pool->name, page->vaddr);
> + list_del(&page->page_list);
> + kfree(page);
Hmm. The is_page_busy case is really a should not happen case.
What is the benefit of skipping the dma_free_coherent and leaking
memory here, vs letting KASAN and friends see the free and possibly
help with debugging? In other words, why is this not:
WARN_ON_ONCE(is_page_busy(page));
dma_free_coherent(pool->dev, pool->allocation, page->vaddr,
page->dma);
...
> page->in_use--;
> *(int *)vaddr = page->offset;
> page->offset = offset;
> - /*
> - * Resist a temptation to do
> - * if (!is_page_busy(page)) pool_free_page(pool, page);
> - * Better have a few empty pages hang around.
> - */
This doesn't look related to the rest, or am I missing something?
Powered by blists - more mailing lists