Skip to contentSkip to conversation

Zillions of warnings in -next

DISCUSSION Alan Cox 5 participants 8 messages Oct 27, 2009 – Oct 28, 2009 Network drivers

lore
Share & tools
lore

Alan Cox

author
#0 lore raw

I was thinking "what a lot of new warnings". Then I dug a little deeper and it seems that someone never bothered to try and build drivers/net/vbus-enet.c on a 32bit system.

So would you mind fixing the 27 new warnings introduced ?

Alan

#1 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line

On 10/27/2009 at 11:25 AM, in message

<[email protected]>, Alan Cox

Quoted 7 lines · <[email protected]> wrote:

I was thinking "what a lot of new warnings". Then I dug a little deeper and it seems that someone never bothered to try and build drivers/net/vbus-enet.c on a 32bit system.

So would you mind fixing the 27 new warnings introduced ?

Alan

Im on it.

Thanks Alan,
-Greg

#2 lore raw
In reply to Alan Cox
Quoted 1 line
Quoted 1 line
Quoted 1 line

On 10/27/2009 at 11:25 AM, in message

<[email protected]>, Alan Cox

Quoted 7 lines · <[email protected]> wrote:

I was thinking "what a lot of new warnings". Then I dug a little deeper and it seems that someone never bothered to try and build drivers/net/vbus-enet.c on a 32bit system.

So would you mind fixing the 27 new warnings introduced ?

Alan

I just did a 32-bit build and indeed reproduced the warnings. However, the warnings appear to be dubious as the code is properly (afaict) annotated with explicit casts, and I believe they are doing what I intended.

Basically I have structures that are always 64 bit (so we can have a mixed-mode 32-bit guest talk to a 64-bit hypervisor, for instance). Therefore I am casting between native and u64, but the compiler doesn't like this. Does anyone have any suggestions on ways to fix this so the compiler is happy?

For instance:

struct foo {
u64 bar;
}

xmit_foo(void *ptr)
{
struct foo f;

f->bar = (u64)ptr; /* generates warning */
}

Kind Regards,
-Greg

author
#3 lore raw
Quoted 8 lines

I just did a 32-bit build and indeed reproduced the warnings. However, the warnings appear to be dubious as the code is properly (afaict) annotated with explicit casts, and I believe they are doing what I intended.

Basically I have structures that are always 64 bit (so we can have a mixed-mode 32-bit guest talk to a 64-bit hypervisor, for instance). Therefore I am casting between native and u64, but the compiler doesn't like this. Does anyone have any suggestions on ways to fix this so the compiler is happy?

foo = (void *)(unsigned long)x

works for the general case because Linux assumes ptr fits ulong

Similarly in the other direction.

Alan

#4 lore raw
Quoted 9 lines · On Tue, 27 Oct 2009 22:21:12 +0000 Alan Cox wrote:
Quoted 8 lines

I just did a 32-bit build and indeed reproduced the warnings. However, the warnings appear to be dubious as the code is properly (afaict) annotated with explicit casts, and I believe they are doing what I intended.

Basically I have structures that are always 64 bit (so we can have a mixed-mode 32-bit guest talk to a 64-bit hypervisor, for instance). Therefore I am casting between native and u64, but the compiler doesn't like this. Does anyone have any suggestions on ways to fix this so the compiler is happy?

Quoted 5 lines

foo = (void *)(unsigned long)x

works for the general case because Linux assumes ptr fits ulong

Similarly in the other direction.

Yes, I already had this done for venet and pci-bridge. Patch below.

---
From: Randy Dunlap <[email protected]>

Eliminate all cast warnings in vbus-enet.c and pci-bridge.c.

---
 drivers/net/vbus-enet.c   |   52 ++++++++++++++++++------------------
 drivers/vbus/pci-bridge.c |   13 ++++-----
 2 files changed, 33 insertions(+), 32 deletions(-)
drivers/net/vbus-enet.c +26 -26
@@ -162,18 +162,18 @@ rxdesc_alloc(struct vbus_enet_priv *priv
 		 * larger than MTU, the host will grab pages out of the
 		 * page-queue and populate additional IOVs
 		 */
-		struct venet_sg *vsg = (struct venet_sg *)desc->cookie;
+		struct venet_sg *vsg = (struct venet_sg *)(unsigned long)desc->cookie;
 		struct venet_iov *iov = &vsg->iov[0];
 
 		memset(vsg, 0, SG_DESC_SIZE);
 
-		vsg->cookie  = (u64)skb;
+		vsg->cookie  = (u64)(unsigned long)skb;
 		vsg->count   = 1;
 
 		iov->ptr     = (u64)__pa(skb->data);
 		iov->len     = len;
 	} else {
-		desc->cookie = (u64)skb;
+		desc->cookie = (u64)(unsigned long)skb;
 		desc->ptr    = (u64)__pa(skb->data);
 		desc->len    = len; /* total length  */
 	}
@@ -207,7 +207,7 @@ rx_pageq_refill(struct vbus_enet_priv *p
 		page = alloc_page(GFP_KERNEL);
 		BUG_ON(!page);
 
-		iter.desc->cookie = (u64)page;
+		iter.desc->cookie = (u64)(unsigned long)page;
 		iter.desc->ptr    = (u64)__pa(page_address(page));
 		iter.desc->len    = PAGE_SIZE;
 
@@ -253,7 +253,7 @@ rx_setup(struct vbus_enet_priv *priv)
 			void *addr = &priv->l4ro.pool[offset];
 
 			iter.desc->ptr    = (u64)offset;
-			iter.desc->cookie = (u64)addr;
+			iter.desc->cookie = (u64)(unsigned long)addr;
 			iter.desc->len    = SG_DESC_SIZE;
 		}
 
@@ -297,19 +297,19 @@ rx_rxq_teardown(struct vbus_enet_priv *p
 			struct venet_sg *vsg;
 			int i;
 
-			vsg = (struct venet_sg *)iter.desc->cookie;
+			vsg = (struct venet_sg *)(unsigned long)iter.desc->cookie;
 
 			/* skip i=0, since that is the skb->data IOV */
 			for (i = 1; i < vsg->count; i++) {
 				struct venet_iov *iov = &vsg->iov[i];
-				struct page *page = (struct page *)iov->ptr;
+				struct page *page = (struct page *)(unsigned long)iov->ptr;
 
 				put_page(page);
 			}
 
-			skb = (struct sk_buff *)vsg->cookie;
+			skb = (struct sk_buff *)(unsigned long)vsg->cookie;
 		} else
-			skb = (struct sk_buff *)iter.desc->cookie;
+			skb = (struct sk_buff *)(unsigned long)iter.desc->cookie;
 
 		iter.desc->valid = 0;
 		wmb();
@@ -341,7 +341,7 @@ rx_l4ro_teardown(struct vbus_enet_priv *
 	 * free each valid descriptor
 	 */
 	while (iter.desc->sown) {
-		struct page *page = (struct page *)iter.desc->cookie;
+		struct page *page = (struct page *)(unsigned long)iter.desc->cookie;
 
 		iter.desc->valid = 0;
 		wmb();
@@ -432,7 +432,7 @@ tx_setup(struct vbus_enet_priv *priv)
 			iter.desc->ptr = (u64)__pa(vsg);
 		}
 
-		iter.desc->cookie = (u64)vsg;
+		iter.desc->cookie = (u64)(unsigned long)vsg;
 		iter.desc->len    = SG_DESC_SIZE;
 
 		ret = ioq_iter_seek(&iter, ioq_seek_next, 0, 0);
@@ -480,7 +480,7 @@ tx_teardown(struct vbus_enet_priv *priv)
 	 * free each valid descriptor
 	 */
 	while (iter.desc->cookie) {
-		struct venet_sg *vsg = (struct venet_sg *)iter.desc->cookie;
+		struct venet_sg *vsg = (struct venet_sg *)(unsigned long)iter.desc->cookie;
 
 		iter.desc->valid = 0;
 		wmb();
@@ -597,8 +597,8 @@ vbus_enet_change_mtu(struct net_device *
 static struct sk_buff *
 vbus_enet_l4ro_import(struct vbus_enet_priv *priv, struct ioq_ring_desc *desc)
 {
-	struct venet_sg *vsg = (struct venet_sg *)desc->cookie;
-	struct sk_buff *skb = (struct sk_buff *)vsg->cookie;
+	struct venet_sg *vsg = (struct venet_sg *)(unsigned long)desc->cookie;
+	struct sk_buff *skb = (struct sk_buff *)(unsigned long)vsg->cookie;
 	struct skb_shared_info *sinfo = skb_shinfo(skb);
 	int i;
 
@@ -618,7 +618,7 @@ vbus_enet_l4ro_import(struct vbus_enet_p
 	/* skip i=0, since that is the skb->data IOV */
 	for (i = 1; i < vsg->count; i++) {
 		struct venet_iov *iov = &vsg->iov[i];
-		struct page *page = (struct page *)iov->ptr;
+		struct page *page = (struct page *)(unsigned long)iov->ptr;
 		skb_frag_t *f = &sinfo->frags[i-1];
 
 		f->page        = page;
@@ -691,7 +691,7 @@ fail:
 static struct sk_buff *
 vbus_enet_flat_import(struct vbus_enet_priv *priv, struct ioq_ring_desc *desc)
 {
-	struct sk_buff *skb = (struct sk_buff *)desc->cookie;
+	struct sk_buff *skb = (struct sk_buff *)(unsigned long)desc->cookie;
 
 	if (!desc->len) {
 		/*
@@ -816,7 +816,7 @@ vbus_enet_tx_start(struct sk_buff *skb, 
 	BUG_ON(iter.desc->sown);
 
 	if (priv->sg) {
-		struct venet_sg *vsg = (struct venet_sg *)iter.desc->cookie;
+		struct venet_sg *vsg = (struct venet_sg *)(unsigned long)iter.desc->cookie;
 		struct scatterlist sgl[MAX_SKB_FRAGS+1];
 		struct scatterlist *sg;
 		int count, maxcount = ARRAY_SIZE(sgl);
@@ -825,7 +825,7 @@ vbus_enet_tx_start(struct sk_buff *skb, 
 
 		memset(vsg, 0, sizeof(*vsg));
 
-		vsg->cookie = (u64)skb;
+		vsg->cookie = (u64)(unsigned long)skb;
 		vsg->len    = skb->len;
 
 		if (skb->ip_summed == CHECKSUM_PARTIAL) {
@@ -873,7 +873,7 @@ vbus_enet_tx_start(struct sk_buff *skb, 
 		 * non scatter-gather mode: simply put the skb right onto the
 		 * ring.
 		 */
-		iter.desc->cookie = (u64)skb;
+		iter.desc->cookie = (u64)(unsigned long)skb;
 		iter.desc->len = (u64)skb->len;
 		iter.desc->ptr = (u64)__pa(skb->data);
 	}
@@ -950,10 +950,10 @@ vbus_enet_tx_reap_one(struct vbus_enet_p
 		if (priv->sg) {
 			struct venet_sg *vsg;
 
-			vsg = (struct venet_sg *)iter.desc->cookie;
-			skb = (struct sk_buff *)vsg->cookie;
+			vsg = (struct venet_sg *)(unsigned long)iter.desc->cookie;
+			skb = (struct sk_buff *)(unsigned long)vsg->cookie;
 		} else
-			skb = (struct sk_buff *)iter.desc->cookie;
+			skb = (struct sk_buff *)(unsigned long)iter.desc->cookie;
 
 		/* Reset the descriptor */
 		iter.desc->valid  = 0;
@@ -1069,7 +1069,7 @@ evq_txc_event(struct vbus_enet_priv *pri
 
 	vbus_enet_tx_reap(priv);
 
-	vbus_enet_skb_complete(priv, (struct sk_buff *)event->cookie);
+	vbus_enet_skb_complete(priv, (struct sk_buff *)(unsigned long)event->cookie);
 }
 
 static void
@@ -1097,7 +1097,7 @@ deferred_evq_isr(unsigned long data)
 	while (!iter.desc->sown) {
 		struct venet_event_header *header;
 
-		header = (struct venet_event_header *)iter.desc->cookie;
+		header = (struct venet_event_header *)(unsigned long)iter.desc->cookie;
 
 		switch (header->id) {
 		case VENET_EVENT_LINKSTATE:
@@ -1112,7 +1112,7 @@ deferred_evq_isr(unsigned long data)
 			break;
 		}
 
-		memset((void *)iter.desc->cookie, 0, priv->evq.evsize);
+		memset((void *)(unsigned long)iter.desc->cookie, 0, priv->evq.evsize);
 
 		/* Advance the in-use tail */
 		ret = ioq_iter_pop(&iter, 0);
@@ -1254,7 +1254,7 @@ vbus_enet_evq_negcap(struct vbus_enet_pr
 			void *addr = &priv->evq.pool[offset];
 
 			iter.desc->ptr    = (u64)offset;
-			iter.desc->cookie = (u64)addr;
+			iter.desc->cookie = (u64)(unsigned long)addr;
 			iter.desc->len    = query.evsize;
 
 			ret = ioq_iter_push(&iter, 0);
drivers/vbus/pci-bridge.c +7 -6
@@ -325,9 +325,10 @@ vbus_pci_device_shm(struct vbus_device_p
 		 */
 		shm_signal_get(&_signal->signal);
 
-		params.signal.offset = (u64)sdesc - (u64)ptr;
+		params.signal.offset = (u64)(unsigned long)sdesc -
+					(u64)(unsigned long)ptr;
 		params.signal.prio   = prio;
-		params.signal.cookie = (u64)_signal;
+		params.signal.cookie = (u64)(unsigned long)_signal;
 
 	} else
 		params.signal.offset = -1; /* yes, this is a u32, but its ok */
@@ -526,7 +527,7 @@ event_devdrop(struct vbus_pci_handle_eve
 static void
 event_shmsignal(struct vbus_pci_handle_event *event)
 {
-	struct _signal *_signal = (struct _signal *)event->handle;
+	struct _signal *_signal = (struct _signal *)(unsigned long)event->handle;
 	struct irq_desc *desc = _signal->desc;
 
 	vbus_pci.stats.notify++;
@@ -536,7 +537,7 @@ event_shmsignal(struct vbus_pci_handle_e
 static void
 event_shmclose(struct vbus_pci_handle_event *event)
 {
-	struct _signal *_signal = (struct _signal *)event->handle;
+	struct _signal *_signal = (struct _signal *)(unsigned long)event->handle;
 
 	/*
 	 * This reference was taken during the DEVICESHM call
@@ -593,7 +594,7 @@ eventq_init(int qlen)
 
 		BUG_ON(iter.desc->valid);
 
-		desc->cookie = (u64)event;
+		desc->cookie = (u64)(unsigned long)event;
 		desc->ptr    = (u64)__pa(event);
 		desc->len    = len; /* total length  */
 		desc->valid  = 1;
@@ -643,7 +644,7 @@ eventq_wakeup(struct ioq_notifier *notif
 		struct ioq_ring_desc *desc  = iter.desc;
 		struct vbus_pci_event *event;
 
-		event = (struct vbus_pci_event *)desc->cookie;
+		event = (struct vbus_pci_event *)(unsigned long)desc->cookie;
 
 		switch (event->eventid) {
 		case VBUS_PCI_EVENT_DEVADD:
#5 lore raw
Quoted 272 lines · Randy Dunlap
Quoted 9 lines · On Tue, 27 Oct 2009 22:21:12 +0000 Alan Cox wrote:
Quoted 8 lines

I just did a 32-bit build and indeed reproduced the warnings. However, the warnings appear to be dubious as the code is properly (afaict) annotated with explicit casts, and I believe they are doing what I intended.

Basically I have structures that are always 64 bit (so we can have a mixed-mode 32-bit guest talk to a 64-bit hypervisor, for instance). Therefore I am casting between native and u64, but the compiler doesn't like this. Does anyone have any suggestions on ways to fix this so the compiler is happy?

Quoted 5 lines

foo = (void *)(unsigned long)x

works for the general case because Linux assumes ptr fits ulong

Similarly in the other direction.

Yes, I already had this done for venet and pci-bridge. Patch below.

---
From: Randy Dunlap <[email protected]>

Eliminate all cast warnings in vbus-enet.c and pci-bridge.c.

---
 drivers/net/vbus-enet.c   |   52 ++++++++++++++++++------------------
 drivers/vbus/pci-bridge.c |   13 ++++-----
 2 files changed, 33 insertions(+), 32 deletions(-)
drivers/net/vbus-enet.c +26 -26
@@ -162,18 +162,18 @@ rxdesc_alloc(struct vbus_enet_priv *priv
 		 * larger than MTU, the host will grab pages out of the
 		 * page-queue and populate additional IOVs
 		 */
-		struct venet_sg *vsg = (struct venet_sg *)desc->cookie;
+		struct venet_sg *vsg = (struct venet_sg *)(unsigned long)desc->cookie;
 		struct venet_iov *iov = &vsg->iov[0];
 
 		memset(vsg, 0, SG_DESC_SIZE);
 
-		vsg->cookie  = (u64)skb;
+		vsg->cookie  = (u64)(unsigned long)skb;
 		vsg->count   = 1;
 
 		iov->ptr     = (u64)__pa(skb->data);
 		iov->len     = len;
 	} else {
-		desc->cookie = (u64)skb;
+		desc->cookie = (u64)(unsigned long)skb;
 		desc->ptr    = (u64)__pa(skb->data);
 		desc->len    = len; /* total length  */
 	}
@@ -207,7 +207,7 @@ rx_pageq_refill(struct vbus_enet_priv *p
 		page = alloc_page(GFP_KERNEL);
 		BUG_ON(!page);
 
-		iter.desc->cookie = (u64)page;
+		iter.desc->cookie = (u64)(unsigned long)page;
 		iter.desc->ptr    = (u64)__pa(page_address(page));
 		iter.desc->len    = PAGE_SIZE;
 
@@ -253,7 +253,7 @@ rx_setup(struct vbus_enet_priv *priv)
 			void *addr = &priv->l4ro.pool[offset];
 
 			iter.desc->ptr    = (u64)offset;
-			iter.desc->cookie = (u64)addr;
+			iter.desc->cookie = (u64)(unsigned long)addr;
 			iter.desc->len    = SG_DESC_SIZE;
 		}
 
@@ -297,19 +297,19 @@ rx_rxq_teardown(struct vbus_enet_priv *p
 			struct venet_sg *vsg;
 			int i;
 
-			vsg = (struct venet_sg *)iter.desc->cookie;
+			vsg = (struct venet_sg *)(unsigned long)iter.desc->cookie;
 
 			/* skip i=0, since that is the skb->data IOV */
 			for (i = 1; i < vsg->count; i++) {
 				struct venet_iov *iov = &vsg->iov[i];
-				struct page *page = (struct page *)iov->ptr;
+				struct page *page = (struct page *)(unsigned long)iov->ptr;
 
 				put_page(page);
 			}
 
-			skb = (struct sk_buff *)vsg->cookie;
+			skb = (struct sk_buff *)(unsigned long)vsg->cookie;
 		} else
-			skb = (struct sk_buff *)iter.desc->cookie;
+			skb = (struct sk_buff *)(unsigned long)iter.desc->cookie;
 
 		iter.desc->valid = 0;
 		wmb();
@@ -341,7 +341,7 @@ rx_l4ro_teardown(struct vbus_enet_priv *
 	 * free each valid descriptor
 	 */
 	while (iter.desc->sown) {
-		struct page *page = (struct page *)iter.desc->cookie;
+		struct page *page = (struct page *)(unsigned long)iter.desc->cookie;
 
 		iter.desc->valid = 0;
 		wmb();
@@ -432,7 +432,7 @@ tx_setup(struct vbus_enet_priv *priv)
 			iter.desc->ptr = (u64)__pa(vsg);
 		}
 
-		iter.desc->cookie = (u64)vsg;
+		iter.desc->cookie = (u64)(unsigned long)vsg;
 		iter.desc->len    = SG_DESC_SIZE;
 
 		ret = ioq_iter_seek(&iter, ioq_seek_next, 0, 0);
@@ -480,7 +480,7 @@ tx_teardown(struct vbus_enet_priv *priv)
 	 * free each valid descriptor
 	 */
 	while (iter.desc->cookie) {
-		struct venet_sg *vsg = (struct venet_sg *)iter.desc->cookie;
+		struct venet_sg *vsg = (struct venet_sg *)(unsigned long)iter.desc->cookie;
 
 		iter.desc->valid = 0;
 		wmb();
@@ -597,8 +597,8 @@ vbus_enet_change_mtu(struct net_device *
 static struct sk_buff *
 vbus_enet_l4ro_import(struct vbus_enet_priv *priv, struct ioq_ring_desc *desc)
 {
-	struct venet_sg *vsg = (struct venet_sg *)desc->cookie;
-	struct sk_buff *skb = (struct sk_buff *)vsg->cookie;
+	struct venet_sg *vsg = (struct venet_sg *)(unsigned long)desc->cookie;
+	struct sk_buff *skb = (struct sk_buff *)(unsigned long)vsg->cookie;
 	struct skb_shared_info *sinfo = skb_shinfo(skb);
 	int i;
 
@@ -618,7 +618,7 @@ vbus_enet_l4ro_import(struct vbus_enet_p
 	/* skip i=0, since that is the skb->data IOV */
 	for (i = 1; i < vsg->count; i++) {
 		struct venet_iov *iov = &vsg->iov[i];
-		struct page *page = (struct page *)iov->ptr;
+		struct page *page = (struct page *)(unsigned long)iov->ptr;
 		skb_frag_t *f = &sinfo->frags[i-1];
 
 		f->page        = page;
@@ -691,7 +691,7 @@ fail:
 static struct sk_buff *
 vbus_enet_flat_import(struct vbus_enet_priv *priv, struct ioq_ring_desc *desc)
 {
-	struct sk_buff *skb = (struct sk_buff *)desc->cookie;
+	struct sk_buff *skb = (struct sk_buff *)(unsigned long)desc->cookie;
 
 	if (!desc->len) {
 		/*
@@ -816,7 +816,7 @@ vbus_enet_tx_start(struct sk_buff *skb, 
 	BUG_ON(iter.desc->sown);
 
 	if (priv->sg) {
-		struct venet_sg *vsg = (struct venet_sg *)iter.desc->cookie;
+		struct venet_sg *vsg = (struct venet_sg *)(unsigned long)iter.desc->cookie;
 		struct scatterlist sgl[MAX_SKB_FRAGS+1];
 		struct scatterlist *sg;
 		int count, maxcount = ARRAY_SIZE(sgl);
@@ -825,7 +825,7 @@ vbus_enet_tx_start(struct sk_buff *skb, 
 
 		memset(vsg, 0, sizeof(*vsg));
 
-		vsg->cookie = (u64)skb;
+		vsg->cookie = (u64)(unsigned long)skb;
 		vsg->len    = skb->len;
 
 		if (skb->ip_summed == CHECKSUM_PARTIAL) {
@@ -873,7 +873,7 @@ vbus_enet_tx_start(struct sk_buff *skb, 
 		 * non scatter-gather mode: simply put the skb right onto the
 		 * ring.
 		 */
-		iter.desc->cookie = (u64)skb;
+		iter.desc->cookie = (u64)(unsigned long)skb;
 		iter.desc->len = (u64)skb->len;
 		iter.desc->ptr = (u64)__pa(skb->data);
 	}
@@ -950,10 +950,10 @@ vbus_enet_tx_reap_one(struct vbus_enet_p
 		if (priv->sg) {
 			struct venet_sg *vsg;
 
-			vsg = (struct venet_sg *)iter.desc->cookie;
-			skb = (struct sk_buff *)vsg->cookie;
+			vsg = (struct venet_sg *)(unsigned long)iter.desc->cookie;
+			skb = (struct sk_buff *)(unsigned long)vsg->cookie;
 		} else
-			skb = (struct sk_buff *)iter.desc->cookie;
+			skb = (struct sk_buff *)(unsigned long)iter.desc->cookie;
 
 		/* Reset the descriptor */
 		iter.desc->valid  = 0;
@@ -1069,7 +1069,7 @@ evq_txc_event(struct vbus_enet_priv *pri
 
 	vbus_enet_tx_reap(priv);
 
-	vbus_enet_skb_complete(priv, (struct sk_buff *)event->cookie);
+	vbus_enet_skb_complete(priv, (struct sk_buff *)(unsigned long)event->cookie);
 }
 
 static void
@@ -1097,7 +1097,7 @@ deferred_evq_isr(unsigned long data)
 	while (!iter.desc->sown) {
 		struct venet_event_header *header;
 
-		header = (struct venet_event_header *)iter.desc->cookie;
+		header = (struct venet_event_header *)(unsigned long)iter.desc->cookie;
 
 		switch (header->id) {
 		case VENET_EVENT_LINKSTATE:
@@ -1112,7 +1112,7 @@ deferred_evq_isr(unsigned long data)
 			break;
 		}
 
-		memset((void *)iter.desc->cookie, 0, priv->evq.evsize);
+		memset((void *)(unsigned long)iter.desc->cookie, 0, priv->evq.evsize);
 
 		/* Advance the in-use tail */
 		ret = ioq_iter_pop(&iter, 0);
@@ -1254,7 +1254,7 @@ vbus_enet_evq_negcap(struct vbus_enet_pr
 			void *addr = &priv->evq.pool[offset];
 
 			iter.desc->ptr    = (u64)offset;
-			iter.desc->cookie = (u64)addr;
+			iter.desc->cookie = (u64)(unsigned long)addr;
 			iter.desc->len    = query.evsize;
 
 			ret = ioq_iter_push(&iter, 0);
drivers/vbus/pci-bridge.c +6 -5
@@ -325,9 +325,10 @@ vbus_pci_device_shm(struct vbus_device_p
 		 */
 		shm_signal_get(&_signal->signal);
 
-		params.signal.offset = (u64)sdesc - (u64)ptr;
+		params.signal.offset = (u64)(unsigned long)sdesc -
+					(u64)(unsigned long)ptr;
 		params.signal.prio   = prio;
-		params.signal.cookie = (u64)_signal;
+		params.signal.cookie = (u64)(unsigned long)_signal;
 
 	} else
 		params.signal.offset = -1; /* yes, this is a u32, but its ok */
@@ -526,7 +527,7 @@ event_devdrop(struct vbus_pci_handle_eve
 static void
 event_shmsignal(struct vbus_pci_handle_event *event)
 {
-	struct _signal *_signal = (struct _signal *)event->handle;
+	struct _signal *_signal = (struct _signal *)(unsigned long)event->handle;
 	struct irq_desc *desc = _signal->desc;
 
 	vbus_pci.stats.notify++;
@@ -536,7 +537,7 @@ event_shmsignal(struct vbus_pci_handle_e
 static void
 event_shmclose(struct vbus_pci_handle_event *event)
 {
-	struct _signal *_signal = (struct _signal *)event->handle;
+	struct _signal *_signal = (struct _signal *)(unsigned long)event->handle;
 
 	/*
 	 * This reference was taken during the DEVICESHM call
@@ -593,7 +594,7 @@ eventq_init(int qlen)
 
 		BUG_ON(iter.desc->valid);
 
-		desc->cookie = (u64)event;
+		desc->cookie = (u64)(unsigned long)event;
 		desc->ptr    = (u64)__pa(event);
 		desc->len    = len; /* total length  */
 		desc->valid  = 1;

drivers/vbus/pci-bridge.c eventq_wakeup(struct ioq_notifier *notif · line 650

Quoted 8 lines
drivers/vbus/pci-bridge.c +1 -1
@@ -643,7 +644,7 @@ eventq_wakeup(struct ioq_notifier *notif
 		struct ioq_ring_desc *desc  = iter.desc;
 		struct vbus_pci_event *event;
 
-		event = (struct vbus_pci_event *)desc->cookie;
+		event = (struct vbus_pci_event *)(unsigned long)desc->cookie;
 
 		switch (event->eventid) {
 		case VBUS_PCI_EVENT_DEVADD:

#define cast_cookie(type, cookie) ((type)(unsigned long)(cookie)) #define assign_cookie(var, cookie) (var) = ((typeof(var))(unsigned long)(cookie))
?

Quoted 4 lines

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#6 lore raw
Quoted 1 line · Joe Perches

On Tue, 2009-10-27 at 15:36 -0700, Randy Dunlap wrote:

[...]

Quoted 7 lines
Quoted 7 lines
drivers/vbus/pci-bridge.c +1 -1
@@ -643,7 +644,7 @@ eventq_wakeup(struct ioq_notifier *notif
 		struct ioq_ring_desc *desc  = iter.desc;
 		struct vbus_pci_event *event;
 
-		event = (struct vbus_pci_event *)desc->cookie;
+		event = (struct vbus_pci_event *)(unsigned long)desc->cookie;
 
Quoted 6 lines

drivers/vbus/pci-bridge.c eventq_wakeup(struct ioq_notifier *notif · line 650

Quoted 2 lines
drivers/vbus/pci-bridge.c +0 -0
 		switch (event->eventid) {
 		case VBUS_PCI_EVENT_DEVADD:

#define cast_cookie(type, cookie) ((type)(unsigned long)(cookie)) #define assign_cookie(var, cookie) (var) = ((typeof(var))(unsigned long)(cookie))

C is usually preferred over cpp. :-)

Signature · 3 lines
Stefan Richter
-=====-==--= =-=- ===--
http://arcgraph.de/sr/
#7 lore raw
In reply to Randy Dunlap
Quoted 10 lines · Randy Dunlap
Quoted 8 lines · On Tue, 27 Oct 2009 22:21:12 +0000 Alan Cox wrote:
Quoted 8 lines

I just did a 32-bit build and indeed reproduced the warnings. However, the warnings appear to be dubious as the code is properly (afaict) annotated with explicit casts, and I believe they are doing what I intended.

Basically I have structures that are always 64 bit (so we can have a mixed-mode 32-bit guest talk to a 64-bit hypervisor, for instance). Therefore I am casting between native and u64, but the compiler doesn't like this. Does anyone have any suggestions on ways to fix this so the compiler is happy?

Quoted 5 lines
Quoted 5 lines

foo = (void *)(unsigned long)x

works for the general case because Linux assumes ptr fits ulong

Similarly in the other direction.

Thank you, Alan. Your tip was spot on.

Quoted 9 lines

Yes, I already had this done for venet and pci-bridge. Patch below.

---
From: Randy Dunlap <[email protected]>

Eliminate all cast warnings in vbus-enet.c and pci-bridge.c.

Thank you, Randy. Applied.

Kind Regards,
-Greg