From: "Fabio M. De Francesco" <fmdefrancesco@gmail.com>
To: netdev@vger.kernel.org,
Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Cc: Ira Weiny <ira.weiny@intel.com>,
Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Subject: Re: [PATCH v2 net-next 3/6] cassini: Use page_address() instead of kmap_atomic()
Date: Fri, 25 Nov 2022 16:38:30 +0100 [thread overview]
Message-ID: <3389596.QJadu78ljV@suse> (raw)
In-Reply-To: <20221123205219.31748-4-anirudh.venkataramanan@intel.com>
On mercoledì 23 novembre 2022 21:52:16 CET Anirudh Venkataramanan wrote:
> Pages for Rx buffers are allocated in cas_page_alloc() using either
> GFP_ATOMIC or GFP_KERNEL. Memory allocated with GFP_KERNEL/GFP_ATOMIC can't
> come from highmem and so there's no need to kmap() them. Just use
> page_address() instead. This makes the variable 'addr' unnecessary, so
> remove it too.
>
> Note that kmap_atomic() disables preemption and page-fault processing,
> but page_address() doesn't. When removing uses of kmap_atomic(), one has to
> check if the code being executed between the map/unmap implicitly depends
> on page-faults and/or preemption being disabled. If yes, then code to
> disable page-faults and/or preemption should also be added for functional
> correctness. That however doesn't appear to be the case here, so just
> page_address() is used.
>
> I don't have hardware, so this change has only been compile tested.
>
> Cc: Ira Weiny <ira.weiny@intel.com>
> Cc: Fabio M. De Francesco <fmdefrancesco@gmail.com>
> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
> ---
> v1 -> v2: Update commit message
> ---
> drivers/net/ethernet/sun/cassini.c | 34 ++++++++++--------------------
> 1 file changed, 11 insertions(+), 23 deletions(-)
Reviewed-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
Thanks,
Fabio
> diff --git a/drivers/net/ethernet/sun/cassini.c
> b/drivers/net/ethernet/sun/cassini.c index 0aca193..2f66cfc 100644
> --- a/drivers/net/ethernet/sun/cassini.c
> +++ b/drivers/net/ethernet/sun/cassini.c
> @@ -1915,7 +1915,7 @@ static int cas_rx_process_pkt(struct cas *cp, struct
> cas_rx_comp *rxc, int off, swivel = RX_SWIVEL_OFF_VAL;
> struct cas_page *page;
> struct sk_buff *skb;
> - void *addr, *crcaddr;
> + void *crcaddr;
> __sum16 csum;
> char *p;
>
> @@ -1936,7 +1936,7 @@ static int cas_rx_process_pkt(struct cas *cp, struct
> cas_rx_comp *rxc, skb_reserve(skb, swivel);
>
> p = skb->data;
> - addr = crcaddr = NULL;
> + crcaddr = NULL;
> if (hlen) { /* always copy header pages */
> i = CAS_VAL(RX_COMP2_HDR_INDEX, words[1]);
> page = cp->rx_pages[CAS_VAL(RX_INDEX_RING, i)]
[CAS_VAL(RX_INDEX_NUM, i)];
> @@ -1948,12 +1948,10 @@ static int cas_rx_process_pkt(struct cas *cp, struct
> cas_rx_comp *rxc, i += cp->crc_size;
> dma_sync_single_for_cpu(&cp->pdev->dev, page->dma_addr +
off,
> i, DMA_FROM_DEVICE);
> - addr = cas_page_map(page->buffer);
> - memcpy(p, addr + off, i);
> + memcpy(p, page_address(page->buffer) + off, i);
> dma_sync_single_for_device(&cp->pdev->dev,
> page->dma_addr + off, i,
> DMA_FROM_DEVICE);
> - cas_page_unmap(addr);
> RX_USED_ADD(page, 0x100);
> p += hlen;
> swivel = 0;
> @@ -1984,12 +1982,11 @@ static int cas_rx_process_pkt(struct cas *cp, struct
> cas_rx_comp *rxc, /* make sure we always copy a header */
> swivel = 0;
> if (p == (char *) skb->data) { /* not split */
> - addr = cas_page_map(page->buffer);
> - memcpy(p, addr + off, RX_COPY_MIN);
> + memcpy(p, page_address(page->buffer) + off,
> + RX_COPY_MIN);
> dma_sync_single_for_device(&cp->pdev->dev,
> page->dma_addr
+ off, i,
>
DMA_FROM_DEVICE);
> - cas_page_unmap(addr);
> off += RX_COPY_MIN;
> swivel = RX_COPY_MIN;
> RX_USED_ADD(page, cp->mtu_stride);
> @@ -2036,10 +2033,8 @@ static int cas_rx_process_pkt(struct cas *cp, struct
> cas_rx_comp *rxc, RX_USED_ADD(page, hlen + cp->crc_size);
> }
>
> - if (cp->crc_size) {
> - addr = cas_page_map(page->buffer);
> - crcaddr = addr + off + hlen;
> - }
> + if (cp->crc_size)
> + crcaddr = page_address(page->buffer) + off +
hlen;
>
> } else {
> /* copying packet */
> @@ -2061,12 +2056,10 @@ static int cas_rx_process_pkt(struct cas *cp, struct
> cas_rx_comp *rxc, i += cp->crc_size;
> dma_sync_single_for_cpu(&cp->pdev->dev, page->dma_addr +
off,
> i, DMA_FROM_DEVICE);
> - addr = cas_page_map(page->buffer);
> - memcpy(p, addr + off, i);
> + memcpy(p, page_address(page->buffer) + off, i);
> dma_sync_single_for_device(&cp->pdev->dev,
> page->dma_addr + off, i,
> DMA_FROM_DEVICE);
> - cas_page_unmap(addr);
> if (p == (char *) skb->data) /* not split */
> RX_USED_ADD(page, cp->mtu_stride);
> else
> @@ -2081,20 +2074,17 @@ static int cas_rx_process_pkt(struct cas *cp, struct
> cas_rx_comp *rxc, page->dma_addr,
> dlen + cp-
>crc_size,
> DMA_FROM_DEVICE);
> - addr = cas_page_map(page->buffer);
> - memcpy(p, addr, dlen + cp->crc_size);
> + memcpy(p, page_address(page->buffer), dlen + cp-
>crc_size);
> dma_sync_single_for_device(&cp->pdev->dev,
> page->dma_addr,
> dlen + cp-
>crc_size,
>
DMA_FROM_DEVICE);
> - cas_page_unmap(addr);
> RX_USED_ADD(page, dlen + cp->crc_size);
> }
> end_copy_pkt:
> - if (cp->crc_size) {
> - addr = NULL;
> + if (cp->crc_size)
> crcaddr = skb->data + alloclen;
> - }
> +
> skb_put(skb, alloclen);
> }
>
> @@ -2103,8 +2093,6 @@ static int cas_rx_process_pkt(struct cas *cp, struct
> cas_rx_comp *rxc, /* checksum includes FCS. strip it out. */
> csum = csum_fold(csum_partial(crcaddr, cp->crc_size,
> csum_unfold(csum)));
> - if (addr)
> - cas_page_unmap(addr);
> }
> skb->protocol = eth_type_trans(skb, cp->dev);
> if (skb->protocol == htons(ETH_P_IP)) {
> --
> 2.37.2
next prev parent reply other threads:[~2022-11-25 15:38 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-11-23 20:52 [PATCH v2 net-next 0/6] Remove uses of kmap_atomic() Anirudh Venkataramanan
2022-11-23 20:52 ` [PATCH v2 net-next 1/6] ch_ktls: Use memcpy_from_page() instead of k[un]map_atomic() Anirudh Venkataramanan
2022-11-24 10:56 ` Ayush Sawal
2022-11-28 20:22 ` Anirudh Venkataramanan
2022-11-25 15:34 ` Fabio M. De Francesco
2022-11-23 20:52 ` [PATCH v2 net-next 2/6] sfc: Use kmap_local_page() instead of kmap_atomic() Anirudh Venkataramanan
2022-11-24 8:34 ` Martin Habets
2022-11-25 15:37 ` Fabio M. De Francesco
2022-11-23 20:52 ` [PATCH v2 net-next 3/6] cassini: Use page_address() " Anirudh Venkataramanan
2022-11-25 15:38 ` Fabio M. De Francesco [this message]
2022-11-23 20:52 ` [PATCH v2 net-next 4/6] cassini: Use memcpy_from_page() instead of k[un]map_atomic() Anirudh Venkataramanan
2022-11-25 15:39 ` Fabio M. De Francesco
2022-11-23 20:52 ` [PATCH v2 net-next 5/6] sunvnet: Use kmap_local_page() instead of kmap_atomic() Anirudh Venkataramanan
2022-11-25 15:40 ` Fabio M. De Francesco
2022-11-23 20:52 ` [PATCH v2 net-next 6/6] net: thunderbolt: " Anirudh Venkataramanan
2022-11-24 9:34 ` Mika Westerberg
2022-11-25 15:41 ` Fabio M. De Francesco
2022-11-25 10:50 ` [PATCH v2 net-next 0/6] Remove uses " patchwork-bot+netdevbpf
2022-11-25 15:47 ` Fabio M. De Francesco
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=3389596.QJadu78ljV@suse \
--to=fmdefrancesco@gmail.com \
--cc=anirudh.venkataramanan@intel.com \
--cc=ira.weiny@intel.com \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).