From: Arend van Spriel <arend@broadcom.com>
To: Kalle Valo <kvalo@codeaurora.org>
Cc: linux-wireless <linux-wireless@vger.kernel.org>,
Hante Meuleman <meuleman@broadcom.com>,
Arend van Spriel <arend@broadcom.com>
Subject: [PATCH 02/14] brcmfmac: prevent possible deadlock on resuming SDIO device.
Date: Sun, 25 Jan 2015 20:31:31 +0100 [thread overview]
Message-ID: <1422214303-7816-3-git-send-email-arend@broadcom.com> (raw)
In-Reply-To: <1422214303-7816-1-git-send-email-arend@broadcom.com>
From: Hante Meuleman <meuleman@broadcom.com>
When the system is resumed a deadlock can occur when DPC gets
entered before resume is complete. This patch fixes this by
properly checking the suspend state outside the claim_host code
block.
Reviewed-by: Arend Van Spriel <arend@broadcom.com>
Reviewed-by: Franky (Zhenhui) Lin <frankyl@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Daniel (Deognyoun) Kim <dekim@broadcom.com>
Signed-off-by: Hante Meuleman <meuleman@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
---
drivers/net/wireless/brcm80211/brcmfmac/bcmsdh.c | 33 ------------------------
drivers/net/wireless/brcm80211/brcmfmac/sdio.c | 18 +++++++++++++
drivers/net/wireless/brcm80211/brcmfmac/sdio.h | 2 --
3 files changed, 18 insertions(+), 35 deletions(-)
diff --git a/drivers/net/wireless/brcm80211/brcmfmac/bcmsdh.c b/drivers/net/wireless/brcm80211/brcmfmac/bcmsdh.c
index 00ba90b8..8ba60f6 100644
--- a/drivers/net/wireless/brcm80211/brcmfmac/bcmsdh.c
+++ b/drivers/net/wireless/brcm80211/brcmfmac/bcmsdh.c
@@ -97,25 +97,6 @@ static void brcmf_sdiod_dummy_irqhandler(struct sdio_func *func)
{
}
-static bool brcmf_sdiod_pm_resume_error(struct brcmf_sdio_dev *sdiodev)
-{
- bool is_err = false;
-#ifdef CONFIG_PM_SLEEP
- is_err = atomic_read(&sdiodev->suspend);
-#endif
- return is_err;
-}
-
-static void brcmf_sdiod_pm_resume_wait(struct brcmf_sdio_dev *sdiodev,
- wait_queue_head_t *wq)
-{
-#ifdef CONFIG_PM_SLEEP
- int retry = 0;
- while (atomic_read(&sdiodev->suspend) && retry++ != 30)
- wait_event_timeout(*wq, false, HZ/100);
-#endif
-}
-
int brcmf_sdiod_intr_register(struct brcmf_sdio_dev *sdiodev)
{
int ret = 0;
@@ -244,10 +225,6 @@ static int brcmf_sdiod_request_data(struct brcmf_sdio_dev *sdiodev, u8 fn,
brcmf_dbg(SDIO, "rw=%d, func=%d, addr=0x%05x, nbytes=%d\n",
write, fn, addr, regsz);
- brcmf_sdiod_pm_resume_wait(sdiodev, &sdiodev->request_word_wait);
- if (brcmf_sdiod_pm_resume_error(sdiodev))
- return -EIO;
-
/* only allow byte access on F0 */
if (WARN_ON(regsz > 1 && !fn))
return -EINVAL;
@@ -462,10 +439,6 @@ static int brcmf_sdiod_buffrw(struct brcmf_sdio_dev *sdiodev, uint fn,
unsigned int req_sz;
int err;
- brcmf_sdiod_pm_resume_wait(sdiodev, &sdiodev->request_buffer_wait);
- if (brcmf_sdiod_pm_resume_error(sdiodev))
- return -EIO;
-
/* Single skb use the standard mmc interface */
req_sz = pkt->len + 3;
req_sz &= (uint)~3;
@@ -516,10 +489,6 @@ static int brcmf_sdiod_sglist_rw(struct brcmf_sdio_dev *sdiodev, uint fn,
if (!pktlist->qlen)
return -EINVAL;
- brcmf_sdiod_pm_resume_wait(sdiodev, &sdiodev->request_buffer_wait);
- if (brcmf_sdiod_pm_resume_error(sdiodev))
- return -EIO;
-
target_list = pktlist;
/* for host with broken sg support, prepare a page aligned list */
__skb_queue_head_init(&local_list);
@@ -1077,8 +1046,6 @@ static int brcmf_ops_sdio_probe(struct sdio_func *func,
#endif
atomic_set(&sdiodev->suspend, false);
- init_waitqueue_head(&sdiodev->request_word_wait);
- init_waitqueue_head(&sdiodev->request_buffer_wait);
brcmf_dbg(SDIO, "F2 found, calling brcmf_sdiod_probe...\n");
err = brcmf_sdiod_probe(sdiodev);
diff --git a/drivers/net/wireless/brcm80211/brcmfmac/sdio.c b/drivers/net/wireless/brcm80211/brcmfmac/sdio.c
index 99a3776..2c4f0cc 100644
--- a/drivers/net/wireless/brcm80211/brcmfmac/sdio.c
+++ b/drivers/net/wireless/brcm80211/brcmfmac/sdio.c
@@ -2609,6 +2609,21 @@ static int brcmf_sdio_intr_rstatus(struct brcmf_sdio *bus)
return ret;
}
+static int brcmf_sdio_pm_resume_wait(struct brcmf_sdio_dev *sdiodev)
+{
+#ifdef CONFIG_PM_SLEEP
+ int retry;
+
+ /* Wait for possible resume to complete */
+ retry = 0;
+ while ((atomic_read(&sdiodev->suspend)) && (retry++ != 50))
+ msleep(20);
+ if (atomic_read(&sdiodev->suspend))
+ return -EIO;
+#endif
+ return 0;
+}
+
static void brcmf_sdio_dpc(struct brcmf_sdio *bus)
{
u32 newstatus = 0;
@@ -2619,6 +2634,9 @@ static void brcmf_sdio_dpc(struct brcmf_sdio *bus)
brcmf_dbg(TRACE, "Enter\n");
+ if (brcmf_sdio_pm_resume_wait(bus->sdiodev))
+ return;
+
sdio_claim_host(bus->sdiodev->func[1]);
/* If waiting for HTAVAIL, check status */
diff --git a/drivers/net/wireless/brcm80211/brcmfmac/sdio.h b/drivers/net/wireless/brcm80211/brcmfmac/sdio.h
index 8eb4262..82494df 100644
--- a/drivers/net/wireless/brcm80211/brcmfmac/sdio.h
+++ b/drivers/net/wireless/brcm80211/brcmfmac/sdio.h
@@ -169,8 +169,6 @@ struct brcmf_sdio_dev {
u32 sbwad; /* Save backplane window address */
struct brcmf_sdio *bus;
atomic_t suspend; /* suspend flag */
- wait_queue_head_t request_word_wait;
- wait_queue_head_t request_buffer_wait;
struct device *dev;
struct brcmf_bus *bus_if;
struct brcmfmac_sdio_platform_data *pdata;
--
1.9.1
next prev parent reply other threads:[~2015-01-25 19:31 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-01-25 19:31 [PATCH 00/14] brcm80211: sdio suspend rework and other fixes Arend van Spriel
2015-01-25 19:31 ` [PATCH 01/14] brcmfmac: Relax scheduling of msgbuf worker on high throughput Arend van Spriel
2015-01-29 8:01 ` [01/14] brcmfmac: Relax scheduling of msgbuf worker on highthroughput Kalle Valo
2015-01-25 19:31 ` Arend van Spriel [this message]
2015-01-25 19:31 ` [PATCH 03/14] brcmfmac: use SDIO DPC for control frames Arend van Spriel
2015-01-25 19:31 ` [PATCH 04/14] brcmfmac: pass DEAUTH/DISASSOC reason code to user-space Arend van Spriel
2015-01-25 19:31 ` [PATCH 05/14] brcmfmac: wait for driver to go idle during suspend Arend van Spriel
2015-01-25 19:31 ` [PATCH 06/14] brcmfmac: SDIO: avoid using bus state for private states Arend van Spriel
2015-01-25 19:31 ` [PATCH 07/14] brcmfmac: Reopen netdev queue on bus state data Arend van Spriel
2015-01-25 19:31 ` [PATCH 08/14] brcmfmac: do not load firmware when device is already running Arend van Spriel
2015-01-25 19:31 ` [PATCH 09/14] brcmutil: use define for boardrev string function Arend van Spriel
2015-01-25 19:31 ` [PATCH 10/14] brcmfmac: determine chip info when not provided by bus layer Arend van Spriel
2015-01-25 19:31 ` [PATCH 11/14] brcmfmac: always obtain device revision info upon intialization Arend van Spriel
2015-01-25 19:31 ` [PATCH 12/14] brcmfmac: show firmware release info in ethtool driver info Arend van Spriel
2015-01-25 19:31 ` [PATCH 13/14] brcmfmac: store revinfo retrieval result Arend van Spriel
2015-01-25 19:31 ` [PATCH 14/14] brcmfmac: fix nvram processing Arend van Spriel
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=1422214303-7816-3-git-send-email-arend@broadcom.com \
--to=arend@broadcom.com \
--cc=kvalo@codeaurora.org \
--cc=linux-wireless@vger.kernel.org \
--cc=meuleman@broadcom.com \
/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).