* Re: [NOT FOR MERGE] ath9k: work around key cache corruption
From: Kalle Valo @ 2016-10-27 6:02 UTC (permalink / raw)
To: Antonio Quartulli; +Cc: ath9k-devel, linux-wireless, sw, Antonio Quartulli
In-Reply-To: <20161026141016.GB26376@prodigo.lan>
Antonio Quartulli <a@unstable.cc> writes:
> On Wed, Oct 26, 2016 at 05:05:14PM +0300, Kalle Valo wrote:
>> Antonio Quartulli <a@unstable.cc> writes:
>>
>> > From: Antonio Quartulli <antonio@open-mesh.com>
>> >
>> > This patch was crafted long time ago to work around a key cache
>> > corruption problem on ath9k chipsets.
>> >
>> > The workaround consists in periodically triggering a worker that
>> > uploads all the keys to the HW cache. The worker is triggered also
>> > when the vif detects 21 undecryptable packets.
>> >
>> >
>> > This patch is based on top compat-wireless-2015-10-26.
>> >
>> >
>> > I was asked to release this code to the public and, since it is
>> > GPL, I am now doing it.
>>
>> GPL? AFAICS ath9k is under ISC. I'm not sure what you mean with the
>> sentence above, but it's possible to interpret it so that this patch is
>> not under ISC license, which is problematic.
>
> Honestly, my sentence was just a way to say "it makes sense to release this
> patch to the public".
I was suspecting it was like that, I just wasn't sure.
> If it needs to be ISC in order to be merged, then it can be released under ISC.
>
> I don't want to enter a legal case :)
Me neither. But I can't apply patches with unclear license.
--
Kalle Valo
^ permalink raw reply
* Re: [PATCH] mwifiex: don't do unbalanced free()'ing in cleanup_if()
From: Dmitry Torokhov @ 2016-10-26 23:52 UTC (permalink / raw)
To: Brian Norris
Cc: Amitkumar Karwar, Nishant Sarmukadam, linux-kernel, Kalle Valo,
linux-wireless, Cathy Luo, Rajat Jain
In-Reply-To: <20161026234353.GA24417@localhost>
On Wed, Oct 26, 2016 at 04:43:54PM -0700, Brian Norris wrote:
> On Wed, Oct 26, 2016 at 04:35:54PM -0700, Dmitry Torokhov wrote:
> > On Wed, Oct 26, 2016 at 04:29:20PM -0700, Brian Norris wrote:
>
> > > diff --git a/drivers/net/wireless/marvell/mwifiex/sdio.c b/drivers/net/wireless/marvell/mwifiex/sdio.c
> > > index 8718950004f3..f04cf5a551b3 100644
> > > --- a/drivers/net/wireless/marvell/mwifiex/sdio.c
> > > +++ b/drivers/net/wireless/marvell/mwifiex/sdio.c
>
> > > @@ -2291,6 +2287,14 @@ static void mwifiex_recreate_adapter(struct sdio_mmc_card *card)
> > >
> > > mwifiex_sdio_remove(func);
> > >
> > > + /*
> > > + * Normally, we would let the driver core take care of releasing these.
> > > + * But we're not letting the driver core handle this one. See above
> > > + * TODO.
> > > + */
> > > + sdio_set_drvdata(func, NULL);
> > > + devm_kfree(&func->dev, card);
> >
> > Ugh, this really messes the unwind order... I guess it is OK since it is
> > the only resource allocated with devm, but I'd be happier if we could
> > reuse existing "card" structure.
>
> I'm really not interested in cleaning up the hacky reset function here
> (see the other TODOs here). I'm sure it's broken in other ways too. In
> its current "design" (if you can call it that) where we remove and
> re-probe the device, I'm not sure there's a way to get it to reuse the
> 'card'.
Ah, I see now... Nevermind then.
Thanks.
--
Dmitry
^ permalink raw reply
* Re: [PATCH] mwifiex: don't do unbalanced free()'ing in cleanup_if()
From: Brian Norris @ 2016-10-26 23:43 UTC (permalink / raw)
To: Dmitry Torokhov
Cc: Amitkumar Karwar, Nishant Sarmukadam, linux-kernel, Kalle Valo,
linux-wireless, Cathy Luo, Rajat Jain
In-Reply-To: <20161026233554.GF27930@dtor-ws>
On Wed, Oct 26, 2016 at 04:35:54PM -0700, Dmitry Torokhov wrote:
> On Wed, Oct 26, 2016 at 04:29:20PM -0700, Brian Norris wrote:
> > diff --git a/drivers/net/wireless/marvell/mwifiex/sdio.c b/drivers/net/wireless/marvell/mwifiex/sdio.c
> > index 8718950004f3..f04cf5a551b3 100644
> > --- a/drivers/net/wireless/marvell/mwifiex/sdio.c
> > +++ b/drivers/net/wireless/marvell/mwifiex/sdio.c
> > @@ -2291,6 +2287,14 @@ static void mwifiex_recreate_adapter(struct sdio_mmc_card *card)
> >
> > mwifiex_sdio_remove(func);
> >
> > + /*
> > + * Normally, we would let the driver core take care of releasing these.
> > + * But we're not letting the driver core handle this one. See above
> > + * TODO.
> > + */
> > + sdio_set_drvdata(func, NULL);
> > + devm_kfree(&func->dev, card);
>
> Ugh, this really messes the unwind order... I guess it is OK since it is
> the only resource allocated with devm, but I'd be happier if we could
> reuse existing "card" structure.
I'm really not interested in cleaning up the hacky reset function here
(see the other TODOs here). I'm sure it's broken in other ways too. In
its current "design" (if you can call it that) where we remove and
re-probe the device, I'm not sure there's a way to get it to reuse the
'card'.
If you insist on refactoring this to protect the potential future unwind
order (if we use devm more heavily), then I guess I'd have to go back to
manual k{zalloc,free}() for sdio.c.
Brian
^ permalink raw reply
* Re: [PATCH] mwifiex: don't do unbalanced free()'ing in cleanup_if()
From: Dmitry Torokhov @ 2016-10-26 23:35 UTC (permalink / raw)
To: Brian Norris
Cc: Amitkumar Karwar, Nishant Sarmukadam, linux-kernel, Kalle Valo,
linux-wireless, Cathy Luo, Rajat Jain
In-Reply-To: <1477524560-49226-1-git-send-email-briannorris@chromium.org>
On Wed, Oct 26, 2016 at 04:29:20PM -0700, Brian Norris wrote:
> The cleanup_if() callback is the inverse of init_if(). We allocate our
> 'card' interface structure in the probe() function, but we free it in
> cleanup_if(). That gives a few problems:
> (a) we leak this memory if probe() fails before we reach init_if()
> (b) we can't safely utilize 'card' after cleanup_if() -- namely, in
> remove() or suspend(), both of which might race with the cleanup
> paths in our asynchronous FW initialization path
>
> Solution: just use devm_kzalloc(), which will free this structure
> properly when the device is removed -- and drop the set_drvdata(...,
> NULL), since the driver core does this for us. This also removes the
> temptation to use drvdata == NULL as a hack for checking if the device
> has been "cleaned up."
>
> I *do* leave the set_drvdata(..., NULL) for the hacky SDIO
> mwifiex_recreate_adapter(), since the device core won't be able to clear
> that one for us.
>
> Signed-off-by: Brian Norris <briannorris@chromium.org>
> ---
> drivers/net/wireless/marvell/mwifiex/pcie.c | 5 +----
> drivers/net/wireless/marvell/mwifiex/sdio.c | 16 ++++++++++------
> drivers/net/wireless/marvell/mwifiex/usb.c | 7 +------
> 3 files changed, 12 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/net/wireless/marvell/mwifiex/pcie.c b/drivers/net/wireless/marvell/mwifiex/pcie.c
> index 063c707844d3..3047c1ab944a 100644
> --- a/drivers/net/wireless/marvell/mwifiex/pcie.c
> +++ b/drivers/net/wireless/marvell/mwifiex/pcie.c
> @@ -189,7 +189,7 @@ static int mwifiex_pcie_probe(struct pci_dev *pdev,
> pr_debug("info: vendor=0x%4.04X device=0x%4.04X rev=%d\n",
> pdev->vendor, pdev->device, pdev->revision);
>
> - card = kzalloc(sizeof(struct pcie_service_card), GFP_KERNEL);
> + card = devm_kzalloc(&pdev->dev, sizeof(*card), GFP_KERNEL);
> if (!card)
> return -ENOMEM;
>
> @@ -2815,7 +2815,6 @@ static int mwifiex_pcie_init(struct mwifiex_adapter *adapter)
> err_set_dma_mask:
> pci_disable_device(pdev);
> err_enable_dev:
> - pci_set_drvdata(pdev, NULL);
> return ret;
> }
>
> @@ -2849,9 +2848,7 @@ static void mwifiex_pcie_cleanup(struct mwifiex_adapter *adapter)
> pci_disable_device(pdev);
> pci_release_region(pdev, 2);
> pci_release_region(pdev, 0);
> - pci_set_drvdata(pdev, NULL);
> }
> - kfree(card);
> }
>
> static int mwifiex_pcie_request_irq(struct mwifiex_adapter *adapter)
> diff --git a/drivers/net/wireless/marvell/mwifiex/sdio.c b/drivers/net/wireless/marvell/mwifiex/sdio.c
> index 8718950004f3..f04cf5a551b3 100644
> --- a/drivers/net/wireless/marvell/mwifiex/sdio.c
> +++ b/drivers/net/wireless/marvell/mwifiex/sdio.c
> @@ -152,7 +152,7 @@ mwifiex_sdio_probe(struct sdio_func *func, const struct sdio_device_id *id)
> pr_debug("info: vendor=0x%4.04X device=0x%4.04X class=%d function=%d\n",
> func->vendor, func->device, func->class, func->num);
>
> - card = kzalloc(sizeof(struct sdio_mmc_card), GFP_KERNEL);
> + card = devm_kzalloc(&func->dev, sizeof(*card), GFP_KERNEL);
> if (!card)
> return -ENOMEM;
>
> @@ -185,7 +185,7 @@ mwifiex_sdio_probe(struct sdio_func *func, const struct sdio_device_id *id)
>
> if (ret) {
> dev_err(&func->dev, "failed to enable function\n");
> - goto err_free;
> + return ret;
> }
>
> /* device tree node parsing and platform specific configuration*/
> @@ -210,8 +210,6 @@ mwifiex_sdio_probe(struct sdio_func *func, const struct sdio_device_id *id)
> sdio_claim_host(func);
> sdio_disable_func(func);
> sdio_release_host(func);
> -err_free:
> - kfree(card);
>
> return ret;
> }
> @@ -2240,8 +2238,6 @@ static void mwifiex_cleanup_sdio(struct mwifiex_adapter *adapter)
> kfree(card->mpa_rx.len_arr);
> kfree(card->mpa_tx.buf);
> kfree(card->mpa_rx.buf);
> - sdio_set_drvdata(card->func, NULL);
> - kfree(card);
> }
>
> /*
> @@ -2291,6 +2287,14 @@ static void mwifiex_recreate_adapter(struct sdio_mmc_card *card)
>
> mwifiex_sdio_remove(func);
>
> + /*
> + * Normally, we would let the driver core take care of releasing these.
> + * But we're not letting the driver core handle this one. See above
> + * TODO.
> + */
> + sdio_set_drvdata(func, NULL);
> + devm_kfree(&func->dev, card);
Ugh, this really messes the unwind order... I guess it is OK since it is
the only resource allocated with devm, but I'd be happier if we could
reuse existing "card" structure.
Thanks.
--
Dmitry
^ permalink raw reply
* [PATCH] mwifiex: don't do unbalanced free()'ing in cleanup_if()
From: Brian Norris @ 2016-10-26 23:29 UTC (permalink / raw)
To: Amitkumar Karwar, Nishant Sarmukadam
Cc: linux-kernel, Kalle Valo, linux-wireless, Cathy Luo, Rajat Jain,
Dmitry Torokhov, Brian Norris
The cleanup_if() callback is the inverse of init_if(). We allocate our
'card' interface structure in the probe() function, but we free it in
cleanup_if(). That gives a few problems:
(a) we leak this memory if probe() fails before we reach init_if()
(b) we can't safely utilize 'card' after cleanup_if() -- namely, in
remove() or suspend(), both of which might race with the cleanup
paths in our asynchronous FW initialization path
Solution: just use devm_kzalloc(), which will free this structure
properly when the device is removed -- and drop the set_drvdata(...,
NULL), since the driver core does this for us. This also removes the
temptation to use drvdata == NULL as a hack for checking if the device
has been "cleaned up."
I *do* leave the set_drvdata(..., NULL) for the hacky SDIO
mwifiex_recreate_adapter(), since the device core won't be able to clear
that one for us.
Signed-off-by: Brian Norris <briannorris@chromium.org>
---
drivers/net/wireless/marvell/mwifiex/pcie.c | 5 +----
drivers/net/wireless/marvell/mwifiex/sdio.c | 16 ++++++++++------
drivers/net/wireless/marvell/mwifiex/usb.c | 7 +------
3 files changed, 12 insertions(+), 16 deletions(-)
diff --git a/drivers/net/wireless/marvell/mwifiex/pcie.c b/drivers/net/wireless/marvell/mwifiex/pcie.c
index 063c707844d3..3047c1ab944a 100644
--- a/drivers/net/wireless/marvell/mwifiex/pcie.c
+++ b/drivers/net/wireless/marvell/mwifiex/pcie.c
@@ -189,7 +189,7 @@ static int mwifiex_pcie_probe(struct pci_dev *pdev,
pr_debug("info: vendor=0x%4.04X device=0x%4.04X rev=%d\n",
pdev->vendor, pdev->device, pdev->revision);
- card = kzalloc(sizeof(struct pcie_service_card), GFP_KERNEL);
+ card = devm_kzalloc(&pdev->dev, sizeof(*card), GFP_KERNEL);
if (!card)
return -ENOMEM;
@@ -2815,7 +2815,6 @@ static int mwifiex_pcie_init(struct mwifiex_adapter *adapter)
err_set_dma_mask:
pci_disable_device(pdev);
err_enable_dev:
- pci_set_drvdata(pdev, NULL);
return ret;
}
@@ -2849,9 +2848,7 @@ static void mwifiex_pcie_cleanup(struct mwifiex_adapter *adapter)
pci_disable_device(pdev);
pci_release_region(pdev, 2);
pci_release_region(pdev, 0);
- pci_set_drvdata(pdev, NULL);
}
- kfree(card);
}
static int mwifiex_pcie_request_irq(struct mwifiex_adapter *adapter)
diff --git a/drivers/net/wireless/marvell/mwifiex/sdio.c b/drivers/net/wireless/marvell/mwifiex/sdio.c
index 8718950004f3..f04cf5a551b3 100644
--- a/drivers/net/wireless/marvell/mwifiex/sdio.c
+++ b/drivers/net/wireless/marvell/mwifiex/sdio.c
@@ -152,7 +152,7 @@ mwifiex_sdio_probe(struct sdio_func *func, const struct sdio_device_id *id)
pr_debug("info: vendor=0x%4.04X device=0x%4.04X class=%d function=%d\n",
func->vendor, func->device, func->class, func->num);
- card = kzalloc(sizeof(struct sdio_mmc_card), GFP_KERNEL);
+ card = devm_kzalloc(&func->dev, sizeof(*card), GFP_KERNEL);
if (!card)
return -ENOMEM;
@@ -185,7 +185,7 @@ mwifiex_sdio_probe(struct sdio_func *func, const struct sdio_device_id *id)
if (ret) {
dev_err(&func->dev, "failed to enable function\n");
- goto err_free;
+ return ret;
}
/* device tree node parsing and platform specific configuration*/
@@ -210,8 +210,6 @@ mwifiex_sdio_probe(struct sdio_func *func, const struct sdio_device_id *id)
sdio_claim_host(func);
sdio_disable_func(func);
sdio_release_host(func);
-err_free:
- kfree(card);
return ret;
}
@@ -2240,8 +2238,6 @@ static void mwifiex_cleanup_sdio(struct mwifiex_adapter *adapter)
kfree(card->mpa_rx.len_arr);
kfree(card->mpa_tx.buf);
kfree(card->mpa_rx.buf);
- sdio_set_drvdata(card->func, NULL);
- kfree(card);
}
/*
@@ -2291,6 +2287,14 @@ static void mwifiex_recreate_adapter(struct sdio_mmc_card *card)
mwifiex_sdio_remove(func);
+ /*
+ * Normally, we would let the driver core take care of releasing these.
+ * But we're not letting the driver core handle this one. See above
+ * TODO.
+ */
+ sdio_set_drvdata(func, NULL);
+ devm_kfree(&func->dev, card);
+
/* power cycle the adapter */
sdio_claim_host(func);
mmc_hw_reset(func->card->host);
diff --git a/drivers/net/wireless/marvell/mwifiex/usb.c b/drivers/net/wireless/marvell/mwifiex/usb.c
index 73eb0846db21..57ed834ba296 100644
--- a/drivers/net/wireless/marvell/mwifiex/usb.c
+++ b/drivers/net/wireless/marvell/mwifiex/usb.c
@@ -382,7 +382,7 @@ static int mwifiex_usb_probe(struct usb_interface *intf,
struct usb_card_rec *card;
u16 id_vendor, id_product, bcd_device, bcd_usb;
- card = kzalloc(sizeof(struct usb_card_rec), GFP_KERNEL);
+ card = devm_kzalloc(&intf->dev, sizeof(*card), GFP_KERNEL);
if (!card)
return -ENOMEM;
@@ -480,7 +480,6 @@ static int mwifiex_usb_probe(struct usb_interface *intf,
if (ret) {
pr_err("%s: mwifiex_add_card failed: %d\n", __func__, ret);
usb_reset_device(udev);
- kfree(card);
return ret;
}
@@ -630,11 +629,7 @@ static void mwifiex_usb_disconnect(struct usb_interface *intf)
"%s: removing card\n", __func__);
mwifiex_remove_card(adapter, &add_remove_card_sem);
- usb_set_intfdata(intf, NULL);
usb_put_dev(interface_to_usbdev(intf));
- kfree(card);
-
- return;
}
static struct usb_driver mwifiex_usb_driver = {
--
2.8.0.rc3.226.g39d4020
^ permalink raw reply related
* [PATCH v2 9/9] mac80211: Claim Fast Initial Link Setup (FILS) STA support
From: Jouni Malinen @ 2016-10-26 21:42 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
In-Reply-To: <1477518126-823-1-git-send-email-jouni@qca.qualcomm.com>
With the previous commits, initial FILS authentication/association
support is now functional in mac80211-based drivers for station role
(and FILS AP case is covered by user space in hostapd withotu requiring
mac80211 changes).
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
---
net/mac80211/main.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/net/mac80211/main.c b/net/mac80211/main.c
index 0d9163c..1822c77 100644
--- a/net/mac80211/main.c
+++ b/net/mac80211/main.c
@@ -549,6 +549,7 @@ struct ieee80211_hw *ieee80211_alloc_hw_nm(size_t priv_data_len,
NL80211_FEATURE_MAC_ON_CREATE |
NL80211_FEATURE_USERSPACE_MPM |
NL80211_FEATURE_FULL_AP_CLIENT_STATE;
+ wiphy_ext_feature_set(wiphy, NL80211_EXT_FEATURE_FILS_STA);
if (!ops->hw_scan)
wiphy->features |= NL80211_FEATURE_LOW_PRIORITY_SCAN |
--
1.9.1
^ permalink raw reply related
* [PATCH v2 8/9] mac80211: FILS AEAD protection for station mode association frames
From: Jouni Malinen @ 2016-10-26 21:42 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
In-Reply-To: <1477518126-823-1-git-send-email-jouni@qca.qualcomm.com>
This adds support for encrypting (Re)Association Request frame and
decryption (Re)Association Response frame when using FILS in station
mode.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
---
net/mac80211/Makefile | 1 +
net/mac80211/aes_cmac.c | 8 +-
net/mac80211/aes_cmac.h | 4 +
net/mac80211/fils_aead.c | 347 +++++++++++++++++++++++++++++++++++++++++++++
net/mac80211/fils_aead.h | 19 +++
net/mac80211/ieee80211_i.h | 4 +
net/mac80211/mlme.c | 27 ++++
7 files changed, 406 insertions(+), 4 deletions(-)
create mode 100644 net/mac80211/fils_aead.c
create mode 100644 net/mac80211/fils_aead.h
diff --git a/net/mac80211/Makefile b/net/mac80211/Makefile
index f9137a8..0b202b3 100644
--- a/net/mac80211/Makefile
+++ b/net/mac80211/Makefile
@@ -19,6 +19,7 @@ mac80211-y := \
aes_gcm.o \
aes_cmac.o \
aes_gmac.o \
+ fils_aead.o \
cfg.o \
ethtool.o \
rx.o \
diff --git a/net/mac80211/aes_cmac.c b/net/mac80211/aes_cmac.c
index bdf0790..d0bd5ff 100644
--- a/net/mac80211/aes_cmac.c
+++ b/net/mac80211/aes_cmac.c
@@ -23,7 +23,7 @@
#define AAD_LEN 20
-static void gf_mulx(u8 *pad)
+void gf_mulx(u8 *pad)
{
int i, carry;
@@ -35,9 +35,9 @@ static void gf_mulx(u8 *pad)
pad[AES_BLOCK_SIZE - 1] ^= 0x87;
}
-static void aes_cmac_vector(struct crypto_cipher *tfm, size_t num_elem,
- const u8 *addr[], const size_t *len, u8 *mac,
- size_t mac_len)
+void aes_cmac_vector(struct crypto_cipher *tfm, size_t num_elem,
+ const u8 *addr[], const size_t *len, u8 *mac,
+ size_t mac_len)
{
u8 cbc[AES_BLOCK_SIZE], pad[AES_BLOCK_SIZE];
const u8 *pos, *end;
diff --git a/net/mac80211/aes_cmac.h b/net/mac80211/aes_cmac.h
index 3702041..c827e1d 100644
--- a/net/mac80211/aes_cmac.h
+++ b/net/mac80211/aes_cmac.h
@@ -11,6 +11,10 @@
#include <linux/crypto.h>
+void gf_mulx(u8 *pad);
+void aes_cmac_vector(struct crypto_cipher *tfm, size_t num_elem,
+ const u8 *addr[], const size_t *len, u8 *mac,
+ size_t mac_len);
struct crypto_cipher *ieee80211_aes_cmac_key_setup(const u8 key[],
size_t key_len);
void ieee80211_aes_cmac(struct crypto_cipher *tfm, const u8 *aad,
diff --git a/net/mac80211/fils_aead.c b/net/mac80211/fils_aead.c
new file mode 100644
index 0000000..afc7efc
--- /dev/null
+++ b/net/mac80211/fils_aead.c
@@ -0,0 +1,347 @@
+/*
+ * FILS AEAD for (Re)Association Request/Response frames
+ * Copyright 2016, Qualcomm Atheros, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <crypto/aes.h>
+#include <crypto/algapi.h>
+#include <crypto/skcipher.h>
+
+#include "ieee80211_i.h"
+#include "aes_cmac.h"
+#include "fils_aead.h"
+
+static int aes_s2v(struct crypto_cipher *tfm,
+ size_t num_elem, const u8 *addr[], size_t len[], u8 *v)
+{
+ u8 d[AES_BLOCK_SIZE], tmp[AES_BLOCK_SIZE];
+ size_t i;
+ const u8 *data[2];
+ size_t data_len[2], data_elems;
+
+ /* D = AES-CMAC(K, <zero>) */
+ memset(tmp, 0, AES_BLOCK_SIZE);
+ data[0] = tmp;
+ data_len[0] = AES_BLOCK_SIZE;
+ aes_cmac_vector(tfm, 1, data, data_len, d, AES_BLOCK_SIZE);
+
+ for (i = 0; i < num_elem - 1; i++) {
+ /* D = dbl(D) xor AES_CMAC(K, Si) */
+ gf_mulx(d); /* dbl */
+ aes_cmac_vector(tfm, 1, &addr[i], &len[i], tmp,
+ AES_BLOCK_SIZE);
+ crypto_xor(d, tmp, AES_BLOCK_SIZE);
+ }
+
+ if (len[i] >= AES_BLOCK_SIZE) {
+ /* len(Sn) >= 128 */
+ size_t j;
+ const u8 *pos;
+
+ /* T = Sn xorend D */
+
+ /* Use a temporary buffer to perform xorend on Sn (addr[i]) to
+ * avoid modifying the const input argument.
+ */
+ data[0] = addr[i];
+ data_len[0] = len[i] - AES_BLOCK_SIZE;
+ pos = addr[i] + data_len[0];
+ for (j = 0; j < AES_BLOCK_SIZE; j++)
+ tmp[j] = pos[j] ^ d[j];
+ data[1] = tmp;
+ data_len[1] = AES_BLOCK_SIZE;
+ data_elems = 2;
+ } else {
+ /* len(Sn) < 128 */
+ /* T = dbl(D) xor pad(Sn) */
+ gf_mulx(d); /* dbl */
+ memset(tmp, 0, AES_BLOCK_SIZE);
+ memcpy(tmp, addr[i], len[i]);
+ tmp[len[i]] = 0x80;
+ crypto_xor(d, tmp, AES_BLOCK_SIZE);
+ data[0] = d;
+ data_len[0] = sizeof(d);
+ data_elems = 1;
+ }
+ /* V = AES-CMAC(K, T) */
+ aes_cmac_vector(tfm, data_elems, data, data_len, v, AES_BLOCK_SIZE);
+
+ return 0;
+}
+
+/* Note: addr[] and len[] needs to have one extra slot at the end. */
+static int aes_siv_encrypt(const u8 *key, size_t key_len,
+ const u8 *plain, size_t plain_len,
+ size_t num_elem, const u8 *addr[],
+ size_t len[], u8 *out)
+{
+ u8 v[AES_BLOCK_SIZE];
+ struct crypto_cipher *tfm;
+ struct crypto_skcipher *tfm2;
+ struct skcipher_request *req;
+ int res;
+ struct scatterlist src[1], dst[1];
+ u8 *tmp;
+
+ key_len /= 2; /* S2V key || CTR key */
+
+ addr[num_elem] = plain;
+ len[num_elem] = plain_len;
+ num_elem++;
+
+ /* S2V */
+
+ tfm = crypto_alloc_cipher("aes", 0, 0);
+ if (IS_ERR(tfm))
+ return PTR_ERR(tfm);
+ /* K1 for S2V */
+ res = crypto_cipher_setkey(tfm, key, key_len);
+ if (!res)
+ res = aes_s2v(tfm, num_elem, addr, len, v);
+ crypto_free_cipher(tfm);
+ if (res)
+ return res;
+
+ /* Use a temporary buffer of the plaintext to handle need for
+ * overwriting this during AES-CTR.
+ */
+ tmp = kmalloc(plain_len, GFP_KERNEL);
+ if (!tmp) {
+ res = -ENOMEM;
+ goto fail;
+ }
+ memcpy(tmp, plain, plain_len);
+
+ /* IV for CTR before encrypted data */
+ memcpy(out, v, AES_BLOCK_SIZE);
+
+ /* Synthetic IV to be used as the initial counter in CTR:
+ * Q = V bitand (1^64 || 0^1 || 1^31 || 0^1 || 1^31)
+ */
+ v[8] &= 0x7f;
+ v[12] &= 0x7f;
+
+ /* CTR */
+
+ tfm2 = crypto_alloc_skcipher("ctr(aes)", 0, 0);
+ if (IS_ERR(tfm2)) {
+ kfree(tmp);
+ return PTR_ERR(tfm2);
+ }
+ /* K2 for CTR */
+ res = crypto_skcipher_setkey(tfm2, key + key_len, key_len);
+ if (res)
+ goto fail;
+
+ req = skcipher_request_alloc(tfm2, GFP_KERNEL);
+ if (!req) {
+ res = -ENOMEM;
+ goto fail;
+ }
+
+ sg_set_buf(&src[0], tmp, plain_len);
+ sg_set_buf(&dst[0], out + AES_BLOCK_SIZE, plain_len);
+ skcipher_request_set_crypt(req, src, dst, plain_len, v);
+ res = crypto_skcipher_encrypt(req);
+ skcipher_request_free(req);
+fail:
+ kfree(tmp);
+ crypto_free_skcipher(tfm2);
+ return res;
+}
+
+/* Note: addr[] and len[] needs to have one extra slot at the end. */
+static int aes_siv_decrypt(const u8 *key, size_t key_len,
+ const u8 *iv_crypt, size_t iv_c_len,
+ size_t num_elem, const u8 *addr[], size_t len[],
+ u8 *out)
+{
+ struct crypto_cipher *tfm;
+ struct crypto_skcipher *tfm2;
+ struct skcipher_request *req;
+ struct scatterlist src[1], dst[1];
+ size_t crypt_len;
+ int res;
+ u8 frame_iv[AES_BLOCK_SIZE], iv[AES_BLOCK_SIZE];
+ u8 check[AES_BLOCK_SIZE];
+
+ crypt_len = iv_c_len - AES_BLOCK_SIZE;
+ key_len /= 2; /* S2V key || CTR key */
+ addr[num_elem] = out;
+ len[num_elem] = crypt_len;
+ num_elem++;
+
+ memcpy(iv, iv_crypt, AES_BLOCK_SIZE);
+ memcpy(frame_iv, iv_crypt, AES_BLOCK_SIZE);
+
+ /* Synthetic IV to be used as the initial counter in CTR:
+ * Q = V bitand (1^64 || 0^1 || 1^31 || 0^1 || 1^31)
+ */
+ iv[8] &= 0x7f;
+ iv[12] &= 0x7f;
+
+ /* CTR */
+
+ tfm2 = crypto_alloc_skcipher("ctr(aes)", 0, 0);
+ if (IS_ERR(tfm2))
+ return PTR_ERR(tfm2);
+ /* K2 for CTR */
+ res = crypto_skcipher_setkey(tfm2, key + key_len, key_len);
+ if (res) {
+ crypto_free_skcipher(tfm2);
+ return res;
+ }
+
+ req = skcipher_request_alloc(tfm2, GFP_KERNEL);
+ if (!req) {
+ crypto_free_skcipher(tfm2);
+ return -ENOMEM;
+ }
+
+ sg_set_buf(&src[0], iv_crypt + AES_BLOCK_SIZE, crypt_len);
+ sg_set_buf(&dst[0], out, crypt_len);
+ skcipher_request_set_crypt(req, src, dst, crypt_len, iv);
+ res = crypto_skcipher_decrypt(req);
+ skcipher_request_free(req);
+ crypto_free_skcipher(tfm2);
+ if (res)
+ return res;
+
+ /* S2V */
+
+ tfm = crypto_alloc_cipher("aes", 0, 0);
+ if (IS_ERR(tfm))
+ return PTR_ERR(tfm);
+ /* K1 for S2V */
+ res = crypto_cipher_setkey(tfm, key, key_len);
+ if (!res)
+ res = aes_s2v(tfm, num_elem, addr, len, check);
+ crypto_free_cipher(tfm);
+ if (res)
+ return res;
+ if (memcmp(check, frame_iv, AES_BLOCK_SIZE) != 0)
+ return -EINVAL;
+ return 0;
+}
+
+int fils_encrypt_assoc_req(struct sk_buff *skb,
+ struct ieee80211_mgd_assoc_data *assoc_data)
+{
+ struct ieee80211_mgmt *mgmt;
+ u8 *capab, *ies, *encr;
+ const u8 *addr[5 + 1], *session;
+ size_t len[5 + 1];
+ size_t crypt_len;
+
+ mgmt = (struct ieee80211_mgmt *)skb->data;
+ if (ieee80211_is_reassoc_req(mgmt->frame_control)) {
+ capab = (u8 *)&mgmt->u.reassoc_req.capab_info;
+ ies = mgmt->u.reassoc_req.variable;
+ } else {
+ capab = (u8 *)&mgmt->u.assoc_req.capab_info;
+ ies = mgmt->u.assoc_req.variable;
+ }
+
+ session = cfg80211_find_ext_ie(WLAN_EID_EXT_FILS_SESSION,
+ ies, skb->data + skb->len - ies);
+ if (!session)
+ return -EINVAL;
+ /* encrypt after FILS Session element */
+ encr = (u8 *)session + 2 + 1 + 8;
+
+ /* AES-SIV AAD vectors */
+
+ /* The STA's MAC address */
+ addr[0] = mgmt->sa;
+ len[0] = ETH_ALEN;
+ /* The AP's BSSID */
+ addr[1] = mgmt->da;
+ len[1] = ETH_ALEN;
+ /* The STA's nonce */
+ addr[2] = assoc_data->fils_nonces;
+ len[2] = FILS_NONCE_LEN;
+ /* The AP's nonce */
+ addr[3] = &assoc_data->fils_nonces[FILS_NONCE_LEN];
+ len[3] = FILS_NONCE_LEN;
+ /* The (Re)Association Request frame from the Capability Information
+ * field to the FILS Session element (both inclusive).
+ */
+ addr[4] = capab;
+ len[4] = encr - capab;
+
+ crypt_len = skb->data + skb->len - encr;
+ skb_put(skb, AES_BLOCK_SIZE);
+ return aes_siv_encrypt(assoc_data->fils_kek, assoc_data->fils_kek_len,
+ encr, crypt_len, 1, addr, len, encr);
+}
+
+int fils_decrypt_assoc_resp(struct ieee80211_sub_if_data *sdata,
+ u8 *frame, size_t *frame_len,
+ struct ieee80211_mgd_assoc_data *assoc_data)
+{
+ struct ieee80211_mgmt *mgmt;
+ u8 *capab, *ies, *encr;
+ const u8 *addr[5 + 1], *session;
+ size_t len[5 + 1];
+ int res;
+ size_t crypt_len;
+
+ if (*frame_len < 24 + 6)
+ return -EINVAL;
+
+ mgmt = (struct ieee80211_mgmt *)frame;
+ capab = (u8 *)&mgmt->u.assoc_resp.capab_info;
+ ies = mgmt->u.assoc_resp.variable;
+ session = cfg80211_find_ext_ie(WLAN_EID_EXT_FILS_SESSION,
+ ies, frame + *frame_len - ies);
+ if (!session) {
+ mlme_dbg(sdata,
+ "No FILS Session element in (Re)Association Response frame from %pM",
+ mgmt->sa);
+ return -EINVAL;
+ }
+ /* decrypt after FILS Session element */
+ encr = (u8 *)session + 2 + 1 + 8;
+
+ /* AES-SIV AAD vectors */
+
+ /* The AP's BSSID */
+ addr[0] = mgmt->sa;
+ len[0] = ETH_ALEN;
+ /* The STA's MAC address */
+ addr[1] = mgmt->da;
+ len[1] = ETH_ALEN;
+ /* The AP's nonce */
+ addr[2] = &assoc_data->fils_nonces[FILS_NONCE_LEN];
+ len[2] = FILS_NONCE_LEN;
+ /* The STA's nonce */
+ addr[3] = assoc_data->fils_nonces;
+ len[3] = FILS_NONCE_LEN;
+ /* The (Re)Association Response frame from the Capability Information
+ * field to the FILS Session element (both inclusive).
+ */
+ addr[4] = capab;
+ len[4] = encr - capab;
+
+ crypt_len = frame + *frame_len - encr;
+ if (crypt_len < AES_BLOCK_SIZE) {
+ mlme_dbg(sdata,
+ "Not enough room for AES-SIV data after FILS Session element in (Re)Association Response frame from %pM",
+ mgmt->sa);
+ return -EINVAL;
+ }
+ res = aes_siv_decrypt(assoc_data->fils_kek, assoc_data->fils_kek_len,
+ encr, crypt_len, 5, addr, len, encr);
+ if (res != 0) {
+ mlme_dbg(sdata,
+ "AES-SIV decryption of (Re)Association Response frame from %pM failed",
+ mgmt->sa);
+ return res;
+ }
+ *frame_len -= AES_BLOCK_SIZE;
+ return 0;
+}
diff --git a/net/mac80211/fils_aead.h b/net/mac80211/fils_aead.h
new file mode 100644
index 0000000..fbc6523
--- /dev/null
+++ b/net/mac80211/fils_aead.h
@@ -0,0 +1,19 @@
+/*
+ * FILS AEAD for (Re)Association Request/Response frames
+ * Copyright 2016, Qualcomm Atheros, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef FILS_AEAD_H
+#define FILS_AEAD_H
+
+int fils_encrypt_assoc_req(struct sk_buff *skb,
+ struct ieee80211_mgd_assoc_data *assoc_data);
+int fils_decrypt_assoc_resp(struct ieee80211_sub_if_data *sdata,
+ u8 *frame, size_t *frame_len,
+ struct ieee80211_mgd_assoc_data *assoc_data);
+
+#endif /* FILS_AEAD_H */
diff --git a/net/mac80211/ieee80211_i.h b/net/mac80211/ieee80211_i.h
index b4e2b6c..d37a577 100644
--- a/net/mac80211/ieee80211_i.h
+++ b/net/mac80211/ieee80211_i.h
@@ -401,6 +401,10 @@ struct ieee80211_mgd_assoc_data {
struct ieee80211_vht_cap ap_vht_cap;
+ u8 fils_nonces[2 * FILS_NONCE_LEN];
+ u8 fils_kek[FILS_MAX_KEK_LEN];
+ size_t fils_kek_len;
+
size_t ie_len;
u8 ie[];
};
diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index b815f2d..f99f9c5 100644
--- a/net/mac80211/mlme.c
+++ b/net/mac80211/mlme.c
@@ -30,6 +30,7 @@
#include "driver-ops.h"
#include "rate.h"
#include "led.h"
+#include "fils_aead.h"
#define IEEE80211_AUTH_TIMEOUT (HZ / 5)
#define IEEE80211_AUTH_TIMEOUT_LONG (HZ / 2)
@@ -652,6 +653,7 @@ static void ieee80211_send_assoc(struct ieee80211_sub_if_data *sdata)
2 + sizeof(struct ieee80211_ht_cap) + /* HT */
2 + sizeof(struct ieee80211_vht_cap) + /* VHT */
assoc_data->ie_len + /* extra IEs */
+ (assoc_data->fils_kek_len ? 16 /* AES-SIV */ : 0) +
9, /* WMM */
GFP_KERNEL);
if (!skb)
@@ -875,6 +877,12 @@ static void ieee80211_send_assoc(struct ieee80211_sub_if_data *sdata)
memcpy(pos, assoc_data->ie + offset, noffset - offset);
}
+ if (assoc_data->fils_kek_len &&
+ fils_encrypt_assoc_req(skb, assoc_data) < 0) {
+ dev_kfree_skb(skb);
+ return;
+ }
+
drv_mgd_prepare_tx(local, sdata);
IEEE80211_SKB_CB(skb)->flags |= IEEE80211_TX_INTFL_DONT_ENCRYPT;
@@ -3146,6 +3154,10 @@ static void ieee80211_rx_mgmt_assoc_resp(struct ieee80211_sub_if_data *sdata,
reassoc ? "Rea" : "A", mgmt->sa,
capab_info, status_code, (u16)(aid & ~(BIT(15) | BIT(14))));
+ if (assoc_data->fils_kek_len &&
+ fils_decrypt_assoc_resp(sdata, (u8 *)mgmt, &len, assoc_data) < 0)
+ return;
+
pos = mgmt->u.assoc_resp.variable;
ieee802_11_parse_elems(pos, len - (pos - (u8 *) mgmt), false, &elems);
@@ -4706,6 +4718,21 @@ int ieee80211_mgd_assoc(struct ieee80211_sub_if_data *sdata,
assoc_data->ie_len = req->ie_len;
}
+ if (req->fils_kek) {
+ if (req->fils_kek_len > FILS_MAX_KEK_LEN) {
+ err = -EINVAL;
+ goto err_free;
+ }
+ memcpy(assoc_data->fils_kek, req->fils_kek,
+ req->fils_kek_len);
+ assoc_data->fils_kek_len = req->fils_kek_len;
+ }
+
+ if (req->fils_nonces) {
+ memcpy(assoc_data->fils_nonces, req->fils_nonces,
+ 2 * FILS_NONCE_LEN);
+ }
+
assoc_data->bss = req->bss;
if (ifmgd->req_smps == IEEE80211_SMPS_AUTOMATIC) {
--
1.9.1
^ permalink raw reply related
* [PATCH v2 7/9] mac80211: Add FILS auth alg mapping
From: Jouni Malinen @ 2016-10-26 21:42 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
In-Reply-To: <1477518126-823-1-git-send-email-jouni@qca.qualcomm.com>
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
---
net/mac80211/mlme.c | 12 ++++++++++++
1 file changed, 12 insertions(+)
diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index b6222f2..b815f2d 100644
--- a/net/mac80211/mlme.c
+++ b/net/mac80211/mlme.c
@@ -2618,6 +2618,9 @@ static void ieee80211_rx_mgmt_auth(struct ieee80211_sub_if_data *sdata,
case WLAN_AUTH_LEAP:
case WLAN_AUTH_FT:
case WLAN_AUTH_SAE:
+ case WLAN_AUTH_FILS_SK:
+ case WLAN_AUTH_FILS_SK_PFS:
+ case WLAN_AUTH_FILS_PK:
break;
case WLAN_AUTH_SHARED_KEY:
if (ifmgd->auth_data->expected_transaction != 4) {
@@ -4479,6 +4482,15 @@ int ieee80211_mgd_auth(struct ieee80211_sub_if_data *sdata,
case NL80211_AUTHTYPE_SAE:
auth_alg = WLAN_AUTH_SAE;
break;
+ case NL80211_AUTHTYPE_FILS_SK:
+ auth_alg = WLAN_AUTH_FILS_SK;
+ break;
+ case NL80211_AUTHTYPE_FILS_SK_PFS:
+ auth_alg = WLAN_AUTH_FILS_SK_PFS;
+ break;
+ case NL80211_AUTHTYPE_FILS_PK:
+ auth_alg = WLAN_AUTH_FILS_PK;
+ break;
default:
return -EOPNOTSUPP;
}
--
1.9.1
^ permalink raw reply related
* [PATCH v2 6/9] cfg80211: Add KEK/nonces for FILS association frames
From: Jouni Malinen @ 2016-10-26 21:42 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
In-Reply-To: <1477518126-823-1-git-send-email-jouni@qca.qualcomm.com>
The new nl80211 attributes can be used to provide KEK and nonces to
allow the driver to encrypt and decrypt FILS (Re)Association
Request/Response frames in station mode.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
---
include/linux/ieee80211.h | 3 +++
include/net/cfg80211.h | 9 +++++++++
include/uapi/linux/nl80211.h | 8 ++++++++
net/wireless/nl80211.c | 12 ++++++++++++
4 files changed, 32 insertions(+)
diff --git a/include/linux/ieee80211.h b/include/linux/ieee80211.h
index 793a017..fe84932 100644
--- a/include/linux/ieee80211.h
+++ b/include/linux/ieee80211.h
@@ -2096,6 +2096,9 @@ enum ieee80211_key_len {
#define IEEE80211_GCMP_MIC_LEN 16
#define IEEE80211_GCMP_PN_LEN 6
+#define FILS_NONCE_LEN 16
+#define FILS_MAX_KEK_LEN 64
+
/* Public action codes */
enum ieee80211_pub_actioncode {
WLAN_PUB_ACTION_EXT_CHANSW_ANN = 4,
diff --git a/include/net/cfg80211.h b/include/net/cfg80211.h
index e1bc478..083fd6d 100644
--- a/include/net/cfg80211.h
+++ b/include/net/cfg80211.h
@@ -1838,6 +1838,12 @@ enum cfg80211_assoc_req_flags {
* @ht_capa_mask: The bits of ht_capa which are to be used.
* @vht_capa: VHT capability override
* @vht_capa_mask: VHT capability mask indicating which fields to use
+ * @fils_kek: FILS KEK for protecting (Re)Association Request/Response frame or
+ * %NULL if FILS is not used.
+ * @fils_kek_len: Length of fils_kek in octets
+ * @fils_nonces: FILS nonces (part of AAD) for protecting (Re)Association
+ * Request/Response frame or %NULL if FILS is not used. This field starts
+ * with 16 octets of STA Nonce followed by 16 octets of AP Nonce.
*/
struct cfg80211_assoc_request {
struct cfg80211_bss *bss;
@@ -1849,6 +1855,9 @@ struct cfg80211_assoc_request {
struct ieee80211_ht_cap ht_capa;
struct ieee80211_ht_cap ht_capa_mask;
struct ieee80211_vht_cap vht_capa, vht_capa_mask;
+ const u8 *fils_kek;
+ size_t fils_kek_len;
+ const u8 *fils_nonces;
};
/**
diff --git a/include/uapi/linux/nl80211.h b/include/uapi/linux/nl80211.h
index dcf8f6f..efa8a49 100644
--- a/include/uapi/linux/nl80211.h
+++ b/include/uapi/linux/nl80211.h
@@ -1942,6 +1942,11 @@ enum nl80211_commands {
* attribute.
* @NL80211_ATTR_NAN_MATCH: used to report a match. This is a nested attribute.
* See &enum nl80211_nan_match_attributes.
+ * @NL80211_ATTR_FILS_KEK: KEK for FILS (Re)Association Request/Response frame
+ * protection.
+ * @NL80211_ATTR_FILS_NONCES: Nonces (part of AAD) for FILS (Re)Association
+ * Request/Response frame protection. This attribute contains the 16 octet
+ * STA Nonce followed by 16 octets of AP Nonce.
*
* @NUM_NL80211_ATTR: total number of nl80211_attrs available
* @NL80211_ATTR_MAX: highest attribute number currently defined
@@ -2342,6 +2347,9 @@ enum nl80211_attrs {
NL80211_ATTR_NAN_FUNC,
NL80211_ATTR_NAN_MATCH,
+ NL80211_ATTR_FILS_KEK,
+ NL80211_ATTR_FILS_NONCES,
+
/* add attributes here, update the policy in nl80211.c */
__NL80211_ATTR_AFTER_LAST,
diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
index 893e321..ec26261 100644
--- a/net/wireless/nl80211.c
+++ b/net/wireless/nl80211.c
@@ -414,6 +414,9 @@ enum nl80211_multicast_groups {
[NL80211_ATTR_NAN_MASTER_PREF] = { .type = NLA_U8 },
[NL80211_ATTR_NAN_DUAL] = { .type = NLA_U8 },
[NL80211_ATTR_NAN_FUNC] = { .type = NLA_NESTED },
+ [NL80211_ATTR_FILS_KEK] = { .type = NLA_BINARY,
+ .len = FILS_MAX_KEK_LEN },
+ [NL80211_ATTR_FILS_NONCES] = { .len = 2 * FILS_NONCE_LEN },
};
/* policy for the key attributes */
@@ -8032,6 +8035,15 @@ static int nl80211_associate(struct sk_buff *skb, struct genl_info *info)
req.flags |= ASSOC_REQ_USE_RRM;
}
+ if (info->attrs[NL80211_ATTR_FILS_KEK]) {
+ req.fils_kek = nla_data(info->attrs[NL80211_ATTR_FILS_KEK]);
+ req.fils_kek_len = nla_len(info->attrs[NL80211_ATTR_FILS_KEK]);
+ if (!info->attrs[NL80211_ATTR_FILS_NONCES])
+ return -EINVAL;
+ req.fils_nonces =
+ nla_data(info->attrs[NL80211_ATTR_FILS_NONCES]);
+ }
+
err = nl80211_crypto_settings(rdev, info, &req.crypto, 1);
if (!err) {
wdev_lock(dev->ieee80211_ptr);
--
1.9.1
^ permalink raw reply related
* [PATCH v2 5/9] cfg80211: Add Fast Initial Link Setup (FILS) auth algs
From: Jouni Malinen @ 2016-10-26 21:42 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
In-Reply-To: <1477518126-823-1-git-send-email-jouni@qca.qualcomm.com>
This defines authentication algorithms for FILS (IEEE 802.11ai).
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
---
include/linux/ieee80211.h | 3 +++
include/uapi/linux/nl80211.h | 6 ++++++
net/wireless/nl80211.c | 21 +++++++++++++++++++--
3 files changed, 28 insertions(+), 2 deletions(-)
diff --git a/include/linux/ieee80211.h b/include/linux/ieee80211.h
index d428adf..793a017 100644
--- a/include/linux/ieee80211.h
+++ b/include/linux/ieee80211.h
@@ -1576,6 +1576,9 @@ struct ieee80211_vht_operation {
#define WLAN_AUTH_SHARED_KEY 1
#define WLAN_AUTH_FT 2
#define WLAN_AUTH_SAE 3
+#define WLAN_AUTH_FILS_SK 4
+#define WLAN_AUTH_FILS_SK_PFS 5
+#define WLAN_AUTH_FILS_PK 6
#define WLAN_AUTH_LEAP 128
#define WLAN_AUTH_CHALLENGE_LEN 128
diff --git a/include/uapi/linux/nl80211.h b/include/uapi/linux/nl80211.h
index f7e0791..dcf8f6f 100644
--- a/include/uapi/linux/nl80211.h
+++ b/include/uapi/linux/nl80211.h
@@ -3667,6 +3667,9 @@ enum nl80211_bss_status {
* @NL80211_AUTHTYPE_FT: Fast BSS Transition (IEEE 802.11r)
* @NL80211_AUTHTYPE_NETWORK_EAP: Network EAP (some Cisco APs and mainly LEAP)
* @NL80211_AUTHTYPE_SAE: Simultaneous authentication of equals
+ * @NL80211_AUTHTYPE_FILS_SK: Fast Initial Link Setup shared key
+ * @NL80211_AUTHTYPE_FILS_SK_PFS: Fast Initial Link Setup shared key with PFS
+ * @NL80211_AUTHTYPE_FILS_PK: Fast Initial Link Setup public key
* @__NL80211_AUTHTYPE_NUM: internal
* @NL80211_AUTHTYPE_MAX: maximum valid auth algorithm
* @NL80211_AUTHTYPE_AUTOMATIC: determine automatically (if necessary by
@@ -3679,6 +3682,9 @@ enum nl80211_auth_type {
NL80211_AUTHTYPE_FT,
NL80211_AUTHTYPE_NETWORK_EAP,
NL80211_AUTHTYPE_SAE,
+ NL80211_AUTHTYPE_FILS_SK,
+ NL80211_AUTHTYPE_FILS_SK_PFS,
+ NL80211_AUTHTYPE_FILS_PK,
/* keep last */
__NL80211_AUTHTYPE_NUM,
diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
index 1a51bd4..893e321 100644
--- a/net/wireless/nl80211.c
+++ b/net/wireless/nl80211.c
@@ -3777,12 +3777,23 @@ static bool nl80211_valid_auth_type(struct cfg80211_registered_device *rdev,
if (!(rdev->wiphy.features & NL80211_FEATURE_SAE) &&
auth_type == NL80211_AUTHTYPE_SAE)
return false;
+ if (!wiphy_ext_feature_isset(&rdev->wiphy,
+ NL80211_EXT_FEATURE_FILS_STA) &&
+ (auth_type == NL80211_AUTHTYPE_FILS_SK ||
+ auth_type == NL80211_AUTHTYPE_FILS_SK_PFS ||
+ auth_type == NL80211_AUTHTYPE_FILS_PK))
+ return false;
return true;
case NL80211_CMD_CONNECT:
case NL80211_CMD_START_AP:
/* SAE not supported yet */
if (auth_type == NL80211_AUTHTYPE_SAE)
return false;
+ /* FILS not supported yet */
+ if (auth_type == NL80211_AUTHTYPE_FILS_SK ||
+ auth_type == NL80211_AUTHTYPE_FILS_SK_PFS ||
+ auth_type == NL80211_AUTHTYPE_FILS_PK)
+ return false;
return true;
default:
return false;
@@ -7809,12 +7820,18 @@ static int nl80211_authenticate(struct sk_buff *skb, struct genl_info *info)
if (!nl80211_valid_auth_type(rdev, auth_type, NL80211_CMD_AUTHENTICATE))
return -EINVAL;
- if (auth_type == NL80211_AUTHTYPE_SAE &&
+ if ((auth_type == NL80211_AUTHTYPE_SAE ||
+ auth_type == NL80211_AUTHTYPE_FILS_SK ||
+ auth_type == NL80211_AUTHTYPE_FILS_SK_PFS ||
+ auth_type == NL80211_AUTHTYPE_FILS_PK) &&
!info->attrs[NL80211_ATTR_AUTH_DATA])
return -EINVAL;
if (info->attrs[NL80211_ATTR_AUTH_DATA]) {
- if (auth_type != NL80211_AUTHTYPE_SAE)
+ if (auth_type != NL80211_AUTHTYPE_SAE &&
+ auth_type != NL80211_AUTHTYPE_FILS_SK &&
+ auth_type != NL80211_AUTHTYPE_FILS_SK_PFS &&
+ auth_type != NL80211_AUTHTYPE_FILS_PK)
return -EINVAL;
auth_data = nla_data(info->attrs[NL80211_ATTR_AUTH_DATA]);
auth_data_len = nla_len(info->attrs[NL80211_ATTR_AUTH_DATA]);
--
1.9.1
^ permalink raw reply related
* [PATCH v2 4/9] cfg80211: Define IEEE P802.11ai (FILS) information elements
From: Jouni Malinen @ 2016-10-26 21:42 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
In-Reply-To: <1477518126-823-1-git-send-email-jouni@qca.qualcomm.com>
Define the Element IDs and Element ID Extensions from IEEE
P802.11ai/D11.0. In addition, add a new cfg80211_find_ext_ie() wrapper
to make it easier to find information elements that used the Element ID
Extension field.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
---
include/linux/ieee80211.h | 20 ++++++++++++++++++++
include/net/cfg80211.h | 21 +++++++++++++++++++++
2 files changed, 41 insertions(+)
diff --git a/include/linux/ieee80211.h b/include/linux/ieee80211.h
index a80516f..d428adf 100644
--- a/include/linux/ieee80211.h
+++ b/include/linux/ieee80211.h
@@ -1960,6 +1960,26 @@ enum ieee80211_eid {
WLAN_EID_VENDOR_SPECIFIC = 221,
WLAN_EID_QOS_PARAMETER = 222,
+ WLAN_EID_CAG_NUMBER = 237,
+ WLAN_EID_AP_CSN = 239,
+ WLAN_EID_FILS_INDICATION = 240,
+ WLAN_EID_DILS = 241,
+ WLAN_EID_FRAGMENT = 242,
+ WLAN_EID_EXTENSION = 255
+};
+
+/* Element ID Extensions for Element ID 255 */
+enum ieee80211_eid_ext {
+ WLAN_EID_EXT_ASSOC_DELAY_INFO = 1,
+ WLAN_EID_EXT_FILS_REQ_PARAMS = 2,
+ WLAN_EID_EXT_FILS_KEY_CONFIRM = 3,
+ WLAN_EID_EXT_FILS_SESSION = 4,
+ WLAN_EID_EXT_FILS_HLP_CONTAINER = 5,
+ WLAN_EID_EXT_FILS_IP_ADDR_ASSIGN = 6,
+ WLAN_EID_EXT_KEY_DELIVERY = 7,
+ WLAN_EID_EXT_FILS_WRAPPED_DATA = 8,
+ WLAN_EID_EXT_FILS_PUBLIC_KEY = 12,
+ WLAN_EID_EXT_FILS_NONCE = 13,
};
/* Action category code */
diff --git a/include/net/cfg80211.h b/include/net/cfg80211.h
index 843ab8a..e1bc478 100644
--- a/include/net/cfg80211.h
+++ b/include/net/cfg80211.h
@@ -4179,6 +4179,27 @@ static inline const u8 *cfg80211_find_ie(u8 eid, const u8 *ies, int len)
}
/**
+ * cfg80211_find_ext_ie - find information element with EID Extension in data
+ *
+ * @eid: element ID Extension
+ * @ies: data consisting of IEs
+ * @len: length of data
+ *
+ * Return: %NULL if the extended element ID could not be found or if
+ * the element is invalid (claims to be longer than the given
+ * data), or a pointer to the first byte of the requested
+ * element, that is the byte containing the element ID.
+ *
+ * Note: There are no checks on the element length other than
+ * having to fit into the given data.
+ */
+static inline const u8 *cfg80211_find_ext_ie(u8 ext_eid, const u8 *ies, int len)
+{
+ return cfg80211_find_ie_match(WLAN_EID_EXTENSION, ies, len,
+ &ext_eid, 1, 2);
+}
+
+/**
* cfg80211_find_vendor_ie - find vendor specific information element in data
*
* @oui: vendor OUI
--
1.9.1
^ permalink raw reply related
* [PATCH v2 3/9] cfg80211: Add feature flag for Fast Initial Link Setup (FILS) as STA
From: Jouni Malinen @ 2016-10-26 21:42 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
In-Reply-To: <1477518126-823-1-git-send-email-jouni@qca.qualcomm.com>
This defines a feature flag that drivers can use to indicate that they
support FILS authentication/association (IEEE 802.11ai) when using user
space SME (NL80211_CMD_AUTHENTICATE) in station mode.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
---
include/uapi/linux/nl80211.h | 3 +++
1 file changed, 3 insertions(+)
diff --git a/include/uapi/linux/nl80211.h b/include/uapi/linux/nl80211.h
index 6f74714..f7e0791 100644
--- a/include/uapi/linux/nl80211.h
+++ b/include/uapi/linux/nl80211.h
@@ -4645,6 +4645,8 @@ enum nl80211_feature_flags {
* configuration (AP/mesh) with HT rates.
* @NL80211_EXT_FEATURE_BEACON_RATE_VHT: Driver supports beacon rate
* configuration (AP/mesh) with VHT rates.
+ * @NL80211_EXT_FEATURE_FILS_STA: This driver supports Fast Initial Link Setup
+ * with user space SME (NL80211_CMD_AUTHENTICATE) in station mode.
*
* @NUM_NL80211_EXT_FEATURES: number of extended features.
* @MAX_NL80211_EXT_FEATURES: highest extended feature index.
@@ -4659,6 +4661,7 @@ enum nl80211_ext_feature_index {
NL80211_EXT_FEATURE_BEACON_RATE_LEGACY,
NL80211_EXT_FEATURE_BEACON_RATE_HT,
NL80211_EXT_FEATURE_BEACON_RATE_VHT,
+ NL80211_EXT_FEATURE_FILS_STA,
/* add new features before the definition below */
NUM_NL80211_EXT_FEATURES,
--
1.9.1
^ permalink raw reply related
* [PATCH v2 2/9] mac80211: Allow AUTH_DATA to be used for FILS
From: Jouni Malinen @ 2016-10-26 21:41 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
In-Reply-To: <1477518126-823-1-git-send-email-jouni@qca.qualcomm.com>
The special SAE case should be limited only for SAE since the more
generic AUTH_DATA can now be used with other authentication algorithms
as well.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
---
net/mac80211/mlme.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index 32fd295..b6222f2 100644
--- a/net/mac80211/mlme.c
+++ b/net/mac80211/mlme.c
@@ -4491,9 +4491,11 @@ int ieee80211_mgd_auth(struct ieee80211_sub_if_data *sdata,
auth_data->bss = req->bss;
if (req->auth_data_len >= 4) {
- __le16 *pos = (__le16 *) req->auth_data;
- auth_data->sae_trans = le16_to_cpu(pos[0]);
- auth_data->sae_status = le16_to_cpu(pos[1]);
+ if (req->auth_type == NL80211_AUTHTYPE_SAE) {
+ __le16 *pos = (__le16 *) req->auth_data;
+ auth_data->sae_trans = le16_to_cpu(pos[0]);
+ auth_data->sae_status = le16_to_cpu(pos[1]);
+ }
memcpy(auth_data->data, req->auth_data + 4,
req->auth_data_len - 4);
auth_data->data_len += req->auth_data_len - 4;
--
1.9.1
^ permalink raw reply related
* [PATCH v2 1/9] cfg80211: Rename SAE_DATA to more generic AUTH_DATA
From: Jouni Malinen @ 2016-10-26 21:41 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
In-Reply-To: <1477518126-823-1-git-send-email-jouni@qca.qualcomm.com>
This adds defines and nl80211 extensions to allow FILS Authentication to
be implemented similarly to SAE. FILS does not need the special rules
for the Authentication transaction number and Status code fields, but it
does need to add non-IE fields. The previously used
NL80211_ATTR_SAE_DATA can be reused for this to avoid having to
duplicate that implementation. Rename that attribute to more generic
NL80211_ATTR_AUTH_DATA (with backwards compatibility define for
NL80211_SAE_DATA).
Also document the special rules related to the Authentication
transaction number and Status code fiels.
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
---
include/net/cfg80211.h | 10 +++++-----
include/uapi/linux/nl80211.h | 13 ++++++++++---
net/mac80211/mlme.c | 12 ++++++------
net/wireless/core.h | 2 +-
net/wireless/mlme.c | 6 +++---
net/wireless/nl80211.c | 18 +++++++++---------
6 files changed, 34 insertions(+), 27 deletions(-)
diff --git a/include/net/cfg80211.h b/include/net/cfg80211.h
index d1ffbc3..843ab8a 100644
--- a/include/net/cfg80211.h
+++ b/include/net/cfg80211.h
@@ -1785,9 +1785,9 @@ struct cfg80211_bss {
* @key_len: length of WEP key for shared key authentication
* @key_idx: index of WEP key for shared key authentication
* @key: WEP key for shared key authentication
- * @sae_data: Non-IE data to use with SAE or %NULL. This starts with
- * Authentication transaction sequence number field.
- * @sae_data_len: Length of sae_data buffer in octets
+ * @auth_data: IE and non-IE data to use with SAE/FILS or %NULL. This starts
+ * with Authentication transaction sequence number field.
+ * @auth_data_len: Length of auth_data buffer in octets
*/
struct cfg80211_auth_request {
struct cfg80211_bss *bss;
@@ -1796,8 +1796,8 @@ struct cfg80211_auth_request {
enum nl80211_auth_type auth_type;
const u8 *key;
u8 key_len, key_idx;
- const u8 *sae_data;
- size_t sae_data_len;
+ const u8 *auth_data;
+ size_t auth_data_len;
};
/**
diff --git a/include/uapi/linux/nl80211.h b/include/uapi/linux/nl80211.h
index 1362d24..6f74714 100644
--- a/include/uapi/linux/nl80211.h
+++ b/include/uapi/linux/nl80211.h
@@ -1638,8 +1638,14 @@ enum nl80211_commands {
* the connection request from a station. nl80211_connect_failed_reason
* enum has different reasons of connection failure.
*
- * @NL80211_ATTR_SAE_DATA: SAE elements in Authentication frames. This starts
- * with the Authentication transaction sequence number field.
+ * @NL80211_ATTR_AUTH_DATA: Fields and elements in Authentication frames. This
+ * starts with the Authentication transaction sequence number field and is
+ * used with authentication algorithms that need special fields to be added
+ * into the frames (SAE and FILS). Currently, only the SAE cases use the
+ * initial two fields (Authentication transaction sequence number and
+ * Status code). However, those fields are included in the attribute data
+ * for all authentication algorithms to keep the attribute definition
+ * consistent.
*
* @NL80211_ATTR_VHT_CAPABILITY: VHT Capability information element (from
* association request when used with NL80211_CMD_NEW_STATION)
@@ -2195,7 +2201,7 @@ enum nl80211_attrs {
NL80211_ATTR_CONN_FAILED_REASON,
- NL80211_ATTR_SAE_DATA,
+ NL80211_ATTR_AUTH_DATA,
NL80211_ATTR_VHT_CAPABILITY,
@@ -2347,6 +2353,7 @@ enum nl80211_attrs {
#define NL80211_ATTR_SCAN_GENERATION NL80211_ATTR_GENERATION
#define NL80211_ATTR_MESH_PARAMS NL80211_ATTR_MESH_CONFIG
#define NL80211_ATTR_IFACE_SOCKET_OWNER NL80211_ATTR_SOCKET_OWNER
+#define NL80211_ATTR_SAE_DATA NL80211_ATTR_AUTH_DATA
/*
* Allow user space programs to use #ifdef on new attributes by defining them
diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index c8d3a9b..32fd295 100644
--- a/net/mac80211/mlme.c
+++ b/net/mac80211/mlme.c
@@ -4483,20 +4483,20 @@ int ieee80211_mgd_auth(struct ieee80211_sub_if_data *sdata,
return -EOPNOTSUPP;
}
- auth_data = kzalloc(sizeof(*auth_data) + req->sae_data_len +
+ auth_data = kzalloc(sizeof(*auth_data) + req->auth_data_len +
req->ie_len, GFP_KERNEL);
if (!auth_data)
return -ENOMEM;
auth_data->bss = req->bss;
- if (req->sae_data_len >= 4) {
- __le16 *pos = (__le16 *) req->sae_data;
+ if (req->auth_data_len >= 4) {
+ __le16 *pos = (__le16 *) req->auth_data;
auth_data->sae_trans = le16_to_cpu(pos[0]);
auth_data->sae_status = le16_to_cpu(pos[1]);
- memcpy(auth_data->data, req->sae_data + 4,
- req->sae_data_len - 4);
- auth_data->data_len += req->sae_data_len - 4;
+ memcpy(auth_data->data, req->auth_data + 4,
+ req->auth_data_len - 4);
+ auth_data->data_len += req->auth_data_len - 4;
}
if (req->ie && req->ie_len) {
diff --git a/net/wireless/core.h b/net/wireless/core.h
index 21e3188..fb2fcd5 100644
--- a/net/wireless/core.h
+++ b/net/wireless/core.h
@@ -345,7 +345,7 @@ int cfg80211_mlme_auth(struct cfg80211_registered_device *rdev,
const u8 *ssid, int ssid_len,
const u8 *ie, int ie_len,
const u8 *key, int key_len, int key_idx,
- const u8 *sae_data, int sae_data_len);
+ const u8 *auth_data, int auth_data_len);
int cfg80211_mlme_assoc(struct cfg80211_registered_device *rdev,
struct net_device *dev,
struct ieee80211_channel *chan,
diff --git a/net/wireless/mlme.c b/net/wireless/mlme.c
index cbb48e2..bd1f7a1 100644
--- a/net/wireless/mlme.c
+++ b/net/wireless/mlme.c
@@ -204,14 +204,14 @@ int cfg80211_mlme_auth(struct cfg80211_registered_device *rdev,
const u8 *ssid, int ssid_len,
const u8 *ie, int ie_len,
const u8 *key, int key_len, int key_idx,
- const u8 *sae_data, int sae_data_len)
+ const u8 *auth_data, int auth_data_len)
{
struct wireless_dev *wdev = dev->ieee80211_ptr;
struct cfg80211_auth_request req = {
.ie = ie,
.ie_len = ie_len,
- .sae_data = sae_data,
- .sae_data_len = sae_data_len,
+ .auth_data = auth_data,
+ .auth_data_len = auth_data_len,
.auth_type = auth_type,
.key = key,
.key_len = key_len,
diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
index ad49a4c..1a51bd4 100644
--- a/net/wireless/nl80211.c
+++ b/net/wireless/nl80211.c
@@ -357,7 +357,7 @@ enum nl80211_multicast_groups {
[NL80211_ATTR_BG_SCAN_PERIOD] = { .type = NLA_U16 },
[NL80211_ATTR_WDEV] = { .type = NLA_U64 },
[NL80211_ATTR_USER_REG_HINT_TYPE] = { .type = NLA_U32 },
- [NL80211_ATTR_SAE_DATA] = { .type = NLA_BINARY, },
+ [NL80211_ATTR_AUTH_DATA] = { .type = NLA_BINARY, },
[NL80211_ATTR_VHT_CAPABILITY] = { .len = NL80211_VHT_CAPABILITY_LEN },
[NL80211_ATTR_SCAN_FLAGS] = { .type = NLA_U32 },
[NL80211_ATTR_P2P_CTWINDOW] = { .type = NLA_U8 },
@@ -7728,8 +7728,8 @@ static int nl80211_authenticate(struct sk_buff *skb, struct genl_info *info)
struct cfg80211_registered_device *rdev = info->user_ptr[0];
struct net_device *dev = info->user_ptr[1];
struct ieee80211_channel *chan;
- const u8 *bssid, *ssid, *ie = NULL, *sae_data = NULL;
- int err, ssid_len, ie_len = 0, sae_data_len = 0;
+ const u8 *bssid, *ssid, *ie = NULL, *auth_data = NULL;
+ int err, ssid_len, ie_len = 0, auth_data_len = 0;
enum nl80211_auth_type auth_type;
struct key_parse key;
bool local_state_change;
@@ -7810,16 +7810,16 @@ static int nl80211_authenticate(struct sk_buff *skb, struct genl_info *info)
return -EINVAL;
if (auth_type == NL80211_AUTHTYPE_SAE &&
- !info->attrs[NL80211_ATTR_SAE_DATA])
+ !info->attrs[NL80211_ATTR_AUTH_DATA])
return -EINVAL;
- if (info->attrs[NL80211_ATTR_SAE_DATA]) {
+ if (info->attrs[NL80211_ATTR_AUTH_DATA]) {
if (auth_type != NL80211_AUTHTYPE_SAE)
return -EINVAL;
- sae_data = nla_data(info->attrs[NL80211_ATTR_SAE_DATA]);
- sae_data_len = nla_len(info->attrs[NL80211_ATTR_SAE_DATA]);
+ auth_data = nla_data(info->attrs[NL80211_ATTR_AUTH_DATA]);
+ auth_data_len = nla_len(info->attrs[NL80211_ATTR_AUTH_DATA]);
/* need to include at least Auth Transaction and Status Code */
- if (sae_data_len < 4)
+ if (auth_data_len < 4)
return -EINVAL;
}
@@ -7836,7 +7836,7 @@ static int nl80211_authenticate(struct sk_buff *skb, struct genl_info *info)
err = cfg80211_mlme_auth(rdev, dev, chan, auth_type, bssid,
ssid, ssid_len, ie, ie_len,
key.p.key, key.p.key_len, key.idx,
- sae_data, sae_data_len);
+ auth_data, auth_data_len);
wdev_unlock(dev->ieee80211_ptr);
return err;
}
--
1.9.1
^ permalink raw reply related
* [PATCH v2 0/9] cfg80211/mac80211: Fast Initial Link Setup (IEEE 802.11ai)
From: Jouni Malinen @ 2016-10-26 21:41 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless, Jouni Malinen
This series adds support for using mac80211-based drivers with Fast
Initial Link Setup as defined in IEEE 802.11ai (to be published early
next year; no more technical changes are expected at this point). The
fils branch in git://w1.fi/hostap.git includes matching commits for
hostapd and wpa_supplicant to use this functionality and initial set of
mac80211_hwsim test cases for the functionality. Actually, most of the
commits are already in the master branch, i.e., only the changes
depending on the nl80211.h changes from this kernel patchset are waiting
in the fils branch.
This series covers only the FILS authentication/association
functionality from IEEE 802.11ai, i.e., the other changes like scanning
optimizations are not included.
v2: Updates to address comments from Johannes
Jouni Malinen (9):
cfg80211: Rename SAE_DATA to more generic AUTH_DATA
mac80211: Allow AUTH_DATA to be used for FILS
cfg80211: Add feature flag for Fast Initial Link Setup (FILS) as STA
cfg80211: Define IEEE P802.11ai (FILS) information elements
cfg80211: Add Fast Initial Link Setup (FILS) auth algs
cfg80211: Add KEK/nonces for FILS association frames
mac80211: Add FILS auth alg mapping
mac80211: FILS AEAD protection for station mode association frames
mac80211: Claim Fast Initial Link Setup (FILS) STA support
include/linux/ieee80211.h | 26 ++++
include/net/cfg80211.h | 40 ++++-
include/uapi/linux/nl80211.h | 30 +++-
net/mac80211/Makefile | 1 +
net/mac80211/aes_cmac.c | 8 +-
net/mac80211/aes_cmac.h | 4 +
net/mac80211/fils_aead.c | 347 +++++++++++++++++++++++++++++++++++++++++++
net/mac80211/fils_aead.h | 19 +++
net/mac80211/ieee80211_i.h | 4 +
net/mac80211/main.c | 1 +
net/mac80211/mlme.c | 57 ++++++-
net/wireless/core.h | 2 +-
net/wireless/mlme.c | 6 +-
net/wireless/nl80211.c | 51 +++++--
14 files changed, 561 insertions(+), 35 deletions(-)
create mode 100644 net/mac80211/fils_aead.c
create mode 100644 net/mac80211/fils_aead.h
--
1.9.1
^ permalink raw reply
* Re: [PATCH v6] mwifiex: parse device tree node for PCIe
From: Rajat Jain @ 2016-10-26 21:16 UTC (permalink / raw)
To: Brian Norris
Cc: Dmitry Torokhov, linux-wireless, devicetree, Xinming Hu,
Amitkumar Karwar, Brian Norris, Kalle Valo, Rob Herring,
Rajat Jain
In-Reply-To: <20161026210805.GA14423@localhost>
On Wed, Oct 26, 2016 at 2:08 PM, Brian Norris <briannorris@chromium.org> wrote:
> On Wed, Oct 26, 2016 at 02:06:48PM -0700, Dmitry Torokhov wrote:
>> On Wed, Oct 26, 2016 at 01:56:34PM -0700, Brian Norris wrote:
>> > On Wed, Oct 26, 2016 at 01:51:48PM -0700, Rajat Jain wrote:
>> > > On Wed, Oct 26, 2016 at 1:46 PM, Dmitry Torokhov
>> > > <dmitry.torokhov@gmail.com> wrote:
>> > > On Wed, Oct 26, 2016 at 01:17:36PM -0700, Brian Norris wrote:
>> > > Sorry, I just saw this... Why do we need devicetree data for
>> > > discoverable bus (PCI)? How does the driver work on systems that do not
>> > > use DT? Why do we need them to behave differently?
>> > >
>> > > There are a couple of out-of-band GPIO pins from Marvell chip that can
>> > > serve as wake-up pins (wake up the CPU when asserted). The Marvell chip
>> > > has to be told which GPIO pin is to be used as the wake-up pin. The pin to
>> > > be used is system / platform dependent. (On some systems it could be
>> > > GPIO13, on others it could be GPIO14 etc depending on how the marvell chip
>> > > is wired up to the CPU).
>>
>> So wakeup pin is not wired to PCIe WAKE?
>
> Not in our case.
>
>> > There's also calibration data. See "marvell,caldata*" and
>> > "marvell,wakeup-pin" properties. Currently only for SDIO, in
>> > Documentation/devicetree/bindings/net/wireless/marvell-sd8xxx.txt, but
>> > we're adding support for PCIe.
>>
>> How would it all work if I moved the PCIe module from one device to
>> another?
>
> These boards are soldered down, at least in the case I care about.
That is right. Since the out of band wake-up pin is not standard on
the PCIe connector - this feature is for soldered chips only.
>
> Brian
^ permalink raw reply
* Re: [PATCH v6] mwifiex: parse device tree node for PCIe
From: Brian Norris @ 2016-10-26 21:08 UTC (permalink / raw)
To: Dmitry Torokhov
Cc: Rajat Jain, linux-wireless, devicetree, Xinming Hu,
Amitkumar Karwar, Brian Norris, Kalle Valo, Rob Herring,
Rajat Jain
In-Reply-To: <20161026210648.GE3989@dtor-ws>
On Wed, Oct 26, 2016 at 02:06:48PM -0700, Dmitry Torokhov wrote:
> On Wed, Oct 26, 2016 at 01:56:34PM -0700, Brian Norris wrote:
> > On Wed, Oct 26, 2016 at 01:51:48PM -0700, Rajat Jain wrote:
> > > On Wed, Oct 26, 2016 at 1:46 PM, Dmitry Torokhov
> > > <dmitry.torokhov@gmail.com> wrote:
> > > On Wed, Oct 26, 2016 at 01:17:36PM -0700, Brian Norris wrote:
> > > Sorry, I just saw this... Why do we need devicetree data for
> > > discoverable bus (PCI)? How does the driver work on systems that do not
> > > use DT? Why do we need them to behave differently?
> > >
> > > There are a couple of out-of-band GPIO pins from Marvell chip that can
> > > serve as wake-up pins (wake up the CPU when asserted). The Marvell chip
> > > has to be told which GPIO pin is to be used as the wake-up pin. The pin to
> > > be used is system / platform dependent. (On some systems it could be
> > > GPIO13, on others it could be GPIO14 etc depending on how the marvell chip
> > > is wired up to the CPU).
>
> So wakeup pin is not wired to PCIe WAKE?
Not in our case.
> > There's also calibration data. See "marvell,caldata*" and
> > "marvell,wakeup-pin" properties. Currently only for SDIO, in
> > Documentation/devicetree/bindings/net/wireless/marvell-sd8xxx.txt, but
> > we're adding support for PCIe.
>
> How would it all work if I moved the PCIe module from one device to
> another?
These boards are soldered down, at least in the case I care about.
Brian
^ permalink raw reply
* Re: [PATCH v6] mwifiex: parse device tree node for PCIe
From: Dmitry Torokhov @ 2016-10-26 21:06 UTC (permalink / raw)
To: Brian Norris
Cc: Rajat Jain, linux-wireless, devicetree, Xinming Hu,
Amitkumar Karwar, Brian Norris, Kalle Valo, Rob Herring,
Rajat Jain
In-Reply-To: <20161026205634.GA13170@localhost>
On Wed, Oct 26, 2016 at 01:56:34PM -0700, Brian Norris wrote:
> On Wed, Oct 26, 2016 at 01:51:48PM -0700, Rajat Jain wrote:
> > On Wed, Oct 26, 2016 at 1:46 PM, Dmitry Torokhov
> > <dmitry.torokhov@gmail.com> wrote:
> > On Wed, Oct 26, 2016 at 01:17:36PM -0700, Brian Norris wrote:
> > Sorry, I just saw this... Why do we need devicetree data for
> > discoverable bus (PCI)? How does the driver work on systems that do not
> > use DT? Why do we need them to behave differently?
> >
> > There are a couple of out-of-band GPIO pins from Marvell chip that can
> > serve as wake-up pins (wake up the CPU when asserted). The Marvell chip
> > has to be told which GPIO pin is to be used as the wake-up pin. The pin to
> > be used is system / platform dependent. (On some systems it could be
> > GPIO13, on others it could be GPIO14 etc depending on how the marvell chip
> > is wired up to the CPU).
So wakeup pin is not wired to PCIe WAKE?
> There's also calibration data. See "marvell,caldata*" and
> "marvell,wakeup-pin" properties. Currently only for SDIO, in
> Documentation/devicetree/bindings/net/wireless/marvell-sd8xxx.txt, but
> we're adding support for PCIe.
How would it all work if I moved the PCIe module from one device to
another?
Thanks.
--
Dmitry
^ permalink raw reply
* Re: [PATCH 7/8] mac80211: FILS AEAD protection for station mode association frames
From: Malinen, Jouni @ 2016-10-26 21:04 UTC (permalink / raw)
To: Johannes Berg; +Cc: linux-wireless@vger.kernel.org
In-Reply-To: <1477460999.4059.13.camel@sipsolutions.net>
On Wed, Oct 26, 2016 at 07:49:59AM +0200, Johannes Berg wrote:
> > +static u8 *fils_find_session(u8 *pos, u8 *end)
> Hmm. I think we should try to write this in terms of cfg80211_find_ie,
> or perhaps cfg80211_find_ie_match, maybe we need to extend those but
> this won't be the only one using the 255 escape.
>=20
> Perhaps just making the eid passed there be a u16, and extending the
> EID definitions appropriately?
Will do that based on our discussion on the details (a new wrapper
function).
> > + if (!session) {
> > + sdata_info(sdata,
> Should we really print the message this prominently? It seems like an
> error case that we may not want to log at all, in case somebody tries
> to flood us with such frames?
...
> Likewise here. Perhaps make these mlme_dbg() or so instead?
These are pretty useful messages for debugging at least now that FILS is
still so new.. Once it gets more mature and has documented
interoperability between independent implementations, we may consider
removing the messages since they would not really show up during normal
operations and would not help much an actual end user. Anyway, I'll
replace them with mlme_dbg() for now.
> > + if (req->fils_nonces)
> > + assoc_data_len +=3D 2 * FILS_NONCE_LEN;
>=20
> Is this really correct? It seems like a bit of an artifact of initially
> having had the nonces in a variable part of the struct?
>=20
> Or perhaps they have to go into the frame in some place that I missed?
> Please add a comment if so.
Ah, looks like I forgot that there and req->fils_kek_len for that
matter. I had initially thought of adding these as dynamically allocated
components at the end of the buffer, but after seeing how req->ie[] was
used, gave up on that extra complexity and simply used fixed size arrays
since both the KEK and FILS nonces do have a known fixed size and we can
easily "waste" that memory in struct ieee80211_mgd_assoc_data for
non-FILS cases without causing any noticeable impact.
> Also, you're never checking req->fils_nonces_set, so you can get rid of
> that.
Indeed. I'll make the cfg80211 patch enforce that both the KEK and
nonces are set since they are both needed for all FILS cases and remove
fils_nonces_set from mac80211.
--=20
Jouni Malinen PGP id EFC895FA=
^ permalink raw reply
* Re: [PATCH v6] mwifiex: parse device tree node for PCIe
From: Brian Norris @ 2016-10-26 20:56 UTC (permalink / raw)
To: Rajat Jain
Cc: Dmitry Torokhov, linux-wireless, devicetree, Xinming Hu,
Amitkumar Karwar, Brian Norris, Kalle Valo, Rob Herring,
Rajat Jain
In-Reply-To: <CACK8Z6H1ACBo7nDxaXR3Vn9cPdyRqxxzdgYz7vNmojiSvjbW5A@mail.gmail.com>
On Wed, Oct 26, 2016 at 01:51:48PM -0700, Rajat Jain wrote:
> On Wed, Oct 26, 2016 at 1:46 PM, Dmitry Torokhov
> <dmitry.torokhov@gmail.com> wrote:
> On Wed, Oct 26, 2016 at 01:17:36PM -0700, Brian Norris wrote:
> Sorry, I just saw this... Why do we need devicetree data for
> discoverable bus (PCI)? How does the driver work on systems that do not
> use DT? Why do we need them to behave differently?
>
> There are a couple of out-of-band GPIO pins from Marvell chip that can
> serve as wake-up pins (wake up the CPU when asserted). The Marvell chip
> has to be told which GPIO pin is to be used as the wake-up pin. The pin to
> be used is system / platform dependent. (On some systems it could be
> GPIO13, on others it could be GPIO14 etc depending on how the marvell chip
> is wired up to the CPU).
There's also calibration data. See "marvell,caldata*" and
"marvell,wakeup-pin" properties. Currently only for SDIO, in
Documentation/devicetree/bindings/net/wireless/marvell-sd8xxx.txt, but
we're adding support for PCIe.
Brian
^ permalink raw reply
* Re: [PATCH v6] mwifiex: parse device tree node for PCIe
From: Dmitry Torokhov @ 2016-10-26 20:46 UTC (permalink / raw)
To: Brian Norris
Cc: Rajat Jain, linux-wireless, devicetree, Xinming Hu,
Amitkumar Karwar, Brian Norris, Kalle Valo, Rob Herring,
rajatxjain
In-Reply-To: <20161026201735.GA10192@localhost>
On Wed, Oct 26, 2016 at 01:17:36PM -0700, Brian Norris wrote:
> Hi Rajat,
>
> On Fri, Oct 21, 2016 at 02:21:09PM -0700, Rajat Jain wrote:
> > From: Xinming Hu <huxm@marvell.com>
> >
> > This patch derives device tree node from pcie bus layer framework, and
> > fixes a minor memory leak in mwifiex_pcie_probe() (in failure path).
> > Device tree bindings file has been renamed(marvell-sd8xxx.txt ->
> > marvell-8xxx.txt) to accommodate PCIe changes.
> >
> > Signed-off-by: Xinming Hu <huxm@marvell.com>
> > Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
> > Signed-off-by: Rajat Jain <rajatja@google.com>
> > Reviewed-by: Brian Norris <briannorris@chromium.org>
> > ---
> > v2: Included vendor and product IDs in compatible strings for PCIe
> > chipsets(Rob Herring)
> > v3: Patch is created using -M option so that it will only include diff of
> > original and renamed files(Rob Herring)
> > Resend v3: Resending the patch because I missed to include device tree mailing
> > while sending v3.
> > v4: Fix error handling, also move-on even if no device tree node is present.
> > v5: Update commit log to include memory leak, return -EINVAL instead of -1.
>
> I've been working on reworking some bugfixes for this driver, and I
> noticed we have some problems w.r.t. memory leaks, and the "memory leak"
> fix is not actually a fix. See below.
Sorry, I just saw this... Why do we need devicetree data for
discoverable bus (PCI)? How does the driver work on systems that do not
use DT? Why do we need them to behave differently?
Thanks.
--
Dmitry
^ permalink raw reply
* Re: compex wle900vx (ath10k) problem on 4.4.24 / armv7
From: Oliver Zemann @ 2016-10-26 20:25 UTC (permalink / raw)
To: Matthias Klein, Michal Kazior, Jonas Gorski; +Cc: linux-wireless
In-Reply-To: <em1c0906d6-d30a-4f2c-bafb-bfec0d18a58d@nb-mak>
>
>> I (naively) went through pci/pm git log and found the following was
>> applied on 4.7-rc2 (i.e. prior to 4.7 release):
>>
>> commit 006d44e49a259b39947366728d65a873a19aadc0
>> Author: Mika Westerberg <mika.westerberg@linux.intel.com>
>> Date: Thu Jun 2 11:17:15 2016 +0300
>>
>> PCI: Add runtime PM support for PCIe ports
>>
>> From reading the commit log it seems to me like it could be it.
>>
>> ath10k tries to wake up the device during probing before it starts
>> talking to it and it does so through MMIO/PCI config space. If it's
>> not mapped properly then driver will not be able to wake it up and
>> will timeout waiting for it.
>>
>> Can you try cherry-picking it into your 4.4.24 and see if it helps?
>>
>>
> Thanks for the help!
>
> Until now I did not compile a kernel for the board.
> I will give it a try and come back with the results ...
>
>
> Best regards,
> Matthias
>
Any news on that? I am in contact with the support of SolidRun
(clearfog). Unfortunately they will stay at 4.4.x - no upgrade to 4.8
nor 4.9 planned. Unfortunately the commit (006d44e) is incompatible to
4.4. because pci_dev does not know bridge_d3
^ permalink raw reply
* Re: [PATCH v6] mwifiex: parse device tree node for PCIe
From: Brian Norris @ 2016-10-26 20:17 UTC (permalink / raw)
To: Rajat Jain
Cc: linux-wireless, devicetree, Xinming Hu, Amitkumar Karwar,
Brian Norris, Kalle Valo, Rob Herring, rajatxjain,
Dmitry Torokhov
In-Reply-To: <1477084869-15612-1-git-send-email-rajatja@google.com>
Hi Rajat,
On Fri, Oct 21, 2016 at 02:21:09PM -0700, Rajat Jain wrote:
> From: Xinming Hu <huxm@marvell.com>
>
> This patch derives device tree node from pcie bus layer framework, and
> fixes a minor memory leak in mwifiex_pcie_probe() (in failure path).
> Device tree bindings file has been renamed(marvell-sd8xxx.txt ->
> marvell-8xxx.txt) to accommodate PCIe changes.
>
> Signed-off-by: Xinming Hu <huxm@marvell.com>
> Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
> Signed-off-by: Rajat Jain <rajatja@google.com>
> Reviewed-by: Brian Norris <briannorris@chromium.org>
> ---
> v2: Included vendor and product IDs in compatible strings for PCIe
> chipsets(Rob Herring)
> v3: Patch is created using -M option so that it will only include diff of
> original and renamed files(Rob Herring)
> Resend v3: Resending the patch because I missed to include device tree mailing
> while sending v3.
> v4: Fix error handling, also move-on even if no device tree node is present.
> v5: Update commit log to include memory leak, return -EINVAL instead of -1.
I've been working on reworking some bugfixes for this driver, and I
noticed we have some problems w.r.t. memory leaks, and the "memory leak"
fix is not actually a fix. See below.
> v6: Remove an unnecessary error print, fix typo in commit log
>
> .../{marvell-sd8xxx.txt => marvell-8xxx.txt} | 8 +++--
> drivers/net/wireless/marvell/mwifiex/pcie.c | 36 +++++++++++++++++++---
> drivers/net/wireless/marvell/mwifiex/sta_cmd.c | 3 +-
> 3 files changed, 39 insertions(+), 8 deletions(-)
> rename Documentation/devicetree/bindings/net/wireless/{marvell-sd8xxx.txt => marvell-8xxx.txt} (91%)
>
> diff --git a/Documentation/devicetree/bindings/net/wireless/marvell-sd8xxx.txt b/Documentation/devicetree/bindings/net/wireless/marvell-8xxx.txt
> similarity index 91%
> rename from Documentation/devicetree/bindings/net/wireless/marvell-sd8xxx.txt
> rename to Documentation/devicetree/bindings/net/wireless/marvell-8xxx.txt
> index c421aba..dfe5f8e 100644
> --- a/Documentation/devicetree/bindings/net/wireless/marvell-sd8xxx.txt
> +++ b/Documentation/devicetree/bindings/net/wireless/marvell-8xxx.txt
> @@ -1,8 +1,8 @@
> -Marvell 8897/8997 (sd8897/sd8997) SDIO devices
> +Marvell 8897/8997 (sd8897/sd8997/pcie8997) SDIO/PCIE devices
> ------
>
> -This node provides properties for controlling the marvell sdio wireless device.
> -The node is expected to be specified as a child node to the SDIO controller that
> +This node provides properties for controlling the marvell sdio/pcie wireless device.
> +The node is expected to be specified as a child node to the SDIO/PCIE controller that
> connects the device to the system.
>
> Required properties:
> @@ -10,6 +10,8 @@ Required properties:
> - compatible : should be one of the following:
> * "marvell,sd8897"
> * "marvell,sd8997"
> + * "pci11ab,2b42"
> + * "pci1b4b,2b42"
>
> Optional properties:
>
> diff --git a/drivers/net/wireless/marvell/mwifiex/pcie.c b/drivers/net/wireless/marvell/mwifiex/pcie.c
> index 3c3c4f1..f7c84d3 100644
> --- a/drivers/net/wireless/marvell/mwifiex/pcie.c
> +++ b/drivers/net/wireless/marvell/mwifiex/pcie.c
> @@ -37,6 +37,22 @@ static struct mwifiex_if_ops pcie_ops;
>
> static struct semaphore add_remove_card_sem;
>
> +static const struct of_device_id mwifiex_pcie_of_match_table[] = {
> + { .compatible = "pci11ab,2b42" },
> + { .compatible = "pci1b4b,2b42" },
> + { }
> +};
> +
> +static int mwifiex_pcie_probe_of(struct device *dev)
> +{
> + if (!of_match_node(mwifiex_pcie_of_match_table, dev->of_node)) {
> + dev_err(dev, "required compatible string missing\n");
> + return -EINVAL;
> + }
> +
> + return 0;
> +}
> +
> static int
> mwifiex_map_pci_memory(struct mwifiex_adapter *adapter, struct sk_buff *skb,
> size_t size, int flags)
> @@ -178,6 +194,7 @@ static int mwifiex_pcie_probe(struct pci_dev *pdev,
> const struct pci_device_id *ent)
> {
> struct pcie_service_card *card;
> + int ret;
>
> pr_debug("info: vendor=0x%4.04X device=0x%4.04X rev=%d\n",
> pdev->vendor, pdev->device, pdev->revision);
> @@ -199,13 +216,24 @@ static int mwifiex_pcie_probe(struct pci_dev *pdev,
> card->pcie.can_ext_scan = data->can_ext_scan;
> }
>
> - if (mwifiex_add_card(card, &add_remove_card_sem, &pcie_ops,
> - MWIFIEX_PCIE)) {
> - pr_err("%s failed\n", __func__);
> - return -1;
> + /* device tree node parsing and platform specific configuration*/
> + if (pdev->dev.of_node) {
> + ret = mwifiex_pcie_probe_of(&pdev->dev);
> + if (ret)
> + goto err_free;
> }
>
> + ret = mwifiex_add_card(card, &add_remove_card_sem, &pcie_ops,
> + MWIFIEX_PCIE);
> + if (ret) {
> + pr_err("%s failed\n", __func__);
> + goto err_free;
For most error cases in mwifiex_add_card(), we call cleanup_if(), which
currently calls kfree(card). It's currently unbalanced, so there are
*some* cases that leak. But...
> + }
> return 0;
> +
> +err_free:
> + kfree(card);
That means that most of the time, this is actually a double-free. I'd
rather have the leak than the double-free :)
Anyway, I have a patch in the works (as part of some device
init/teardown bugfixes) that will convert the allocation to
devm_kzalloc() and drop the kfree()'ing. So that'll fix the
aforementioned bug.
In your next revision (sorry), please just drop this "leak" fix.
Regards,
Brian
> + return ret;
> }
>
> /*
> diff --git a/drivers/net/wireless/marvell/mwifiex/sta_cmd.c b/drivers/net/wireless/marvell/mwifiex/sta_cmd.c
> index 2a162c3..c8dccf5 100644
> --- a/drivers/net/wireless/marvell/mwifiex/sta_cmd.c
> +++ b/drivers/net/wireless/marvell/mwifiex/sta_cmd.c
> @@ -2218,7 +2218,8 @@ int mwifiex_sta_init_cmd(struct mwifiex_private *priv, u8 first_sta, bool init)
> * The cal-data can be read from device tree and/or
> * a configuration file and downloaded to firmware.
> */
> - if (priv->adapter->iface_type == MWIFIEX_SDIO &&
> + if ((priv->adapter->iface_type == MWIFIEX_SDIO ||
> + priv->adapter->iface_type == MWIFIEX_PCIE) &&
> adapter->dev->of_node) {
> adapter->dt_node = adapter->dev->of_node;
> if (of_property_read_u32(adapter->dt_node,
> --
> 2.8.0.rc3.226.g39d4020
>
^ permalink raw reply
* RE: [PATCH 2/5] mwifiex: use spinlock for 'mwifiex_processing' in shutdown_drv
From: Amitkumar Karwar @ 2016-10-26 16:59 UTC (permalink / raw)
To: Dmitry Torokhov
Cc: Brian Norris, linux-wireless@vger.kernel.org, Cathy Luo,
Nishant Sarmukadam
In-Reply-To: <20161026163607.GA3989@dtor-ws>
Hi Dmitry,
> From: Dmitry Torokhov [mailto:dmitry.torokhov@gmail.com]
> Sent: Wednesday, October 26, 2016 10:06 PM
> To: Amitkumar Karwar
> Cc: Brian Norris; linux-wireless@vger.kernel.org; Cathy Luo; Nishant
> Sarmukadam
> Subject: Re: [PATCH 2/5] mwifiex: use spinlock for 'mwifiex_processing'
> in shutdown_drv
>
> Hi Amit,
>
> On Wed, Oct 26, 2016 at 03:23:08PM +0000, Amitkumar Karwar wrote:
> >
> > This race won't occur. At this point of time(i.e while calling
> mwifiex_shutdown_drv() in deinit), following things are completed. We
> don't expect mwifiex_main_process() to be scheduled.
> > 1) Connection to peer device is terminated at the beginning of
> teardown thread. So we don't receive any Tx data from kernel.
> > 2) Last command SHUTDOWN is exchanged with firmware. So there won't be
> any activity/interrupt from firmware.
> > 3) Interrupts are disabled.
> > 4) "adapter->surprise_removed" flag is set. It will skip
> mwifiex_main_process() calls.
> >
> > -----------
> > static void mwifiex_main_work_queue(struct work_struct *work) {
> > struct mwifiex_adapter *adapter =
> > container_of(work, struct mwifiex_adapter, main_work);
> >
> > if (adapter->surprise_removed)
> > return;
> > mwifiex_main_process(adapter); }
> > ----------
> > 5) We have "mwifiex_terminate_workqueue(adapter)" call to flush and
> destroy workqueue.
>
> OK, but if interrupts are disabled and you ensure that work is flushed
> or completed before you call mwifiex_shutdown_drv() then I do not
> understand why you need all of this at all? Why do you need to check
> status in mwifiex_shutdown_drv() and why do you want
> mwifiex_main_process() to call mwifiex_shutdown_drv() in certain cases?
> Can you simply remove all this stuff?
>
I agree. This code is there for long time. I will prepare a patch for this cleanup work.
Regards,
Amitkumar
^ permalink raw reply
* Re: [PATCH 2/5] mwifiex: use spinlock for 'mwifiex_processing' in shutdown_drv
From: Dmitry Torokhov @ 2016-10-26 16:36 UTC (permalink / raw)
To: Amitkumar Karwar
Cc: Brian Norris, linux-wireless@vger.kernel.org, Cathy Luo,
Nishant Sarmukadam
In-Reply-To: <24731897638e42ba8b05acd6afafef3b@SC-EXCH04.marvell.com>
Hi Amit,
On Wed, Oct 26, 2016 at 03:23:08PM +0000, Amitkumar Karwar wrote:
>
> This race won't occur. At this point of time(i.e while calling mwifiex_shutdown_drv() in deinit), following things are completed. We don't expect mwifiex_main_process() to be scheduled.
> 1) Connection to peer device is terminated at the beginning of teardown thread. So we don't receive any Tx data from kernel.
> 2) Last command SHUTDOWN is exchanged with firmware. So there won't be any activity/interrupt from firmware.
> 3) Interrupts are disabled.
> 4) "adapter->surprise_removed" flag is set. It will skip mwifiex_main_process() calls.
>
> -----------
> static void mwifiex_main_work_queue(struct work_struct *work)
> {
> struct mwifiex_adapter *adapter =
> container_of(work, struct mwifiex_adapter, main_work);
>
> if (adapter->surprise_removed)
> return;
> mwifiex_main_process(adapter);
> }
> ----------
> 5) We have "mwifiex_terminate_workqueue(adapter)" call to flush and destroy workqueue.
OK, but if interrupts are disabled and you ensure that work is flushed
or completed before you call mwifiex_shutdown_drv() then I do not
understand why you need all of this at all? Why do you need to check
status in mwifiex_shutdown_drv() and why do you want
mwifiex_main_process() to call mwifiex_shutdown_drv() in certain cases?
Can you simply remove all this stuff?
Thanks.
--
Dmitry
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox