Skip to contentSkip to conversation

dm: Add support for all targets which requires MANUAL_RELOC

RFC PATCH New Michal Simek 5 participants 19 messages Feb 2, 2015 – Feb 12, 2015

lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
drivers/core/root.c +64 -0
index 47b3acfbe981..1a5287b2fe08 100644
@@ -36,6 +36,65 @@ struct udevice *dm_root(void)
 	return gd->dm_root;
 }
 
+#if defined(CONFIG_NEEDS_MANUAL_RELOC)
+void fix_drivers(void)
+{
+	struct driver *drv =
+		ll_entry_start(struct driver, driver);
+	const int n_ents = ll_entry_count(struct driver, driver);
+	struct driver *entry;
+
+	for (entry = drv; entry != drv + n_ents; entry++) {
+		if (entry->of_match)
+			entry->of_match = (const struct udevice_id *)
+				((u32)entry->of_match + gd->reloc_off);
+		if (entry->bind)
+			entry->bind += gd->reloc_off;
+		if (entry->probe)
+			entry->probe += gd->reloc_off;
+		if (entry->remove)
+			entry->remove += gd->reloc_off;
+		if (entry->unbind)
+			entry->unbind += gd->reloc_off;
+		if (entry->ofdata_to_platdata)
+			entry->ofdata_to_platdata += gd->reloc_off;
+		if (entry->child_pre_probe)
+			entry->child_pre_probe += gd->reloc_off;
+		if (entry->child_post_remove)
+			entry->child_post_remove += gd->reloc_off;
+		/* OPS are fixed in every uclass post_probe function */
+		if (entry->ops)
+			entry->ops += gd->reloc_off;
+	}
+}
+
+void fix_uclass(void)
+{
+	struct uclass_driver *uclass =
+		ll_entry_start(struct uclass_driver, uclass);
+	const int n_ents = ll_entry_count(struct uclass_driver, uclass);
+	struct uclass_driver *entry;
+
+	for (entry = uclass; entry != uclass + n_ents; entry++) {
+		if (entry->post_bind)
+			entry->post_bind += gd->reloc_off;
+		if (entry->pre_unbind)
+			entry->pre_unbind += gd->reloc_off;
+		if (entry->post_probe)
+			entry->post_probe += gd->reloc_off;
+		if (entry->pre_remove)
+			entry->pre_remove += gd->reloc_off;
+		if (entry->init)
+			entry->init += gd->reloc_off;
+		if (entry->destroy)
+			entry->destroy += gd->reloc_off;
+		/* FIXME maybe also need to fix these ops */
+		if (entry->ops)
+			entry->ops += gd->reloc_off;
+	}
+}
+#endif
+
 int dm_init(void)
 {
 	int ret;
@@ -46,6 +105,11 @@ int dm_init(void)
 	}
 	INIT_LIST_HEAD(&DM_UCLASS_ROOT_NON_CONST);
 
+#if defined(CONFIG_NEEDS_MANUAL_RELOC)
+	fix_drivers();
+	fix_uclass();
+#endif
+
 	ret = device_bind_by_name(NULL, false, &root_info, &DM_ROOT_NON_CONST);
 	if (ret)
 		return ret;
drivers/serial/serial-uclass.c +16 -0
index d1b5777cecda..f13484b8539c 100644
@@ -258,6 +258,22 @@ static int serial_post_probe(struct udevice *dev)
 #endif
 	int ret;
 
+#if defined(CONFIG_NEEDS_MANUAL_RELOC)
+	if (ops->setbrg)
+		ops->setbrg += gd->reloc_off;
+	if (ops->getc)
+		ops->getc += gd->reloc_off;
+	if (ops->putc)
+		ops->putc += gd->reloc_off;
+	if (ops->pending)
+		ops->pending += gd->reloc_off;
+	if (ops->clear)
+		ops->clear += gd->reloc_off;
+#if CONFIG_POST & CONFIG_SYS_POST_UART
+	if (ops->loop)
+		ops->loop += gd->reloc_off
+#endif
+#endif
 	/* Set the baud rate */
 	if (ops->setbrg) {
 		ret = ops->setbrg(dev, gd->baudrate);

--
1.8.2.3

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20150202/113bfadb/attachment.sig>

#1 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

Regards,
Simon

#2 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi.

Quoted 11 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 8 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

Quoted 6 lines
Quoted 3 lines
 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

If we use SPL, we do not have to relocate code, I think.

Best Regards
Masahiro Yamada

#3 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 2 lines

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

Thanks,
Michal

#4 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi Michal,

Quoted 39 lines · Michal Simek

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 1 line
Quoted 3 lines
Quoted 1 line

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

Maybe I asked this before, but when can we remove CONFIG_NEEDS_MANUAL_RELOC? What platforms need it?

Regards,
Simon

#5 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi Simon,

Quoted 2 lines

Maybe I asked this before, but when can we remove CONFIG_NEEDS_MANUAL_RELOC? What platforms need it?

A quick grep yields a global define of CONFIG_NEEDS_MANUAL_RELOC for avr32, m68k, nds32, and sparc

Next question is - which of these platforms have toolchains which lack the ability to generate the relocation tables that we need?

Regards,

Graeme

#6 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi,

Quoted 6 lines · Graeme Russ

Hi Simon,

Quoted 2 lines

Maybe I asked this before, but when can we remove CONFIG_NEEDS_MANUAL_RELOC? What platforms need it?

A quick grep yields a global define of CONFIG_NEEDS_MANUAL_RELOC for avr32, m68k, nds32, and sparc

+ Microblaze to move to generic board support now.

Quoted 2 lines

Next question is - which of these platforms have toolchains which lack the ability to generate the relocation tables that we need?

Microblaze have the option to generate relocation tables without need to use CONFIG_NEEDS_MANUAL_RELOC.

Thanks,
Michal

#7 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

In reply to Simon Glass

Hi Michal,

Quoted 42 lines · Simon Glass

Hi Michal,

Quoted 39 lines · Michal Simek

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 1 line
Quoted 6 lines
Quoted 3 lines
Quoted 1 line

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

Maybe I asked this before, but when can we remove CONFIG_NEEDS_MANUAL_RELOC? What platforms need it?

Sorry if you answered these questions but can you please resend if so as I missed it.

Thanks,
Simon

#8 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi,

Quoted 47 lines · Simon Glass

Hi Michal,

Quoted 44 lines · Simon Glass

Hi Michal,

Quoted 41 lines · Michal Simek

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 2 lines

If we use SPL, we do not have to relocate code, I think.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

SPL doesn't have relocation that's why this code is not used there.

Maybe I asked this before, but when can we remove CONFIG_NEEDS_MANUAL_RELOC? What platforms need it?

Sorry if you answered these questions but can you please resend if so as I missed it.

Graeme has answered it + my response here.
http://lists.denx.de/pipermail/u-boot/2015-February/203911.html

removing: When all platforms are moved to REL/RELA we can remove this manual reloc option.

Thanks,
Michal

#9 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Quoted 55 lines · Michal Simek

Hi,

Quoted 52 lines · Simon Glass

Hi Michal,

Quoted 48 lines · Simon Glass

Hi Michal,

Quoted 39 lines · Michal Simek

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 1 line
Quoted 3 lines
Quoted 1 line

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

Maybe I asked this before, but when can we remove CONFIG_NEEDS_MANUAL_RELOC? What platforms need it?

Quoted 8 lines
Quoted 2 lines

Sorry if you answered these questions but can you please resend if so as I missed it.

Graeme has answered it + my response here.
http://lists.denx.de/pipermail/u-boot/2015-February/203911.html

removing: When all platforms are moved to REL/RELA we can remove this manual reloc option.

With a heavy heart:

#10 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi Simon,

Quoted 62 lines · Simon Glass
Quoted 61 lines · Michal Simek

Hi,

Quoted 47 lines · Simon Glass

Hi Michal,

Quoted 44 lines · Simon Glass

Hi Michal,

Quoted 41 lines · Michal Simek

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 2 lines

If we use SPL, we do not have to relocate code, I think.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

SPL doesn't have relocation that's why this code is not used there.

Maybe I asked this before, but when can we remove CONFIG_NEEDS_MANUAL_RELOC? What platforms need it?

Sorry if you answered these questions but can you please resend if so as I missed it.

Graeme has answered it + my response here.
http://lists.denx.de/pipermail/u-boot/2015-February/203911.html

Quoted 7 lines
Quoted 2 lines

removing: When all platforms are moved to REL/RELA we can remove this manual reloc option.

With a heavy heart:

Are you going to take this patch to your DM tree?

Thanks,
Michal

#11 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi MIchal,

Quoted 68 lines · Michal Simek

Hi Simon,

Quoted 65 lines · Simon Glass
Quoted 63 lines · Michal Simek

Hi,

Quoted 47 lines · Simon Glass

Hi Michal,

Quoted 44 lines · Simon Glass

Hi Michal,

Quoted 41 lines · Michal Simek

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 2 lines

If we use SPL, we do not have to relocate code, I think.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

SPL doesn't have relocation that's why this code is not used there.

Maybe I asked this before, but when can we remove CONFIG_NEEDS_MANUAL_RELOC? What platforms need it?

Sorry if you answered these questions but can you please resend if so as I missed it.

Graeme has answered it + my response here.
http://lists.denx.de/pipermail/u-boot/2015-February/203911.html

removing: When all platforms are moved to REL/RELA we can remove this manual reloc option.

Quoted 6 lines
Quoted 4 lines

With a heavy heart:

Are you going to take this patch to your DM tree?

I'm hoping to bring in the Kconfig patches and send a pull request this week, so could do that if it suits. But it is up to you, so let's do whatever makes your life easier. Let me know.

Regards,
Simon

#12 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Quoted 73 lines · Simon Glass

Hi MIchal,

Quoted 70 lines · Michal Simek

Hi Simon,

Quoted 67 lines · Simon Glass
Quoted 55 lines · Michal Simek

Hi,

Quoted 52 lines · Simon Glass

Hi Michal,

Quoted 48 lines · Simon Glass

Hi Michal,

Quoted 39 lines · Michal Simek

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 1 line
Quoted 3 lines
Quoted 1 line

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

Maybe I asked this before, but when can we remove CONFIG_NEEDS_MANUAL_RELOC? What platforms need it?

Quoted 8 lines
Quoted 2 lines

Sorry if you answered these questions but can you please resend if so as I missed it.

Graeme has answered it + my response here.
http://lists.denx.de/pipermail/u-boot/2015-February/203911.html

removing: When all platforms are moved to REL/RELA we can remove this manual reloc option.

With a heavy heart:

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

Are you going to take this patch to your DM tree?

I'm hoping to bring in the Kconfig patches and send a pull request this week, so could do that if it suits. But it is up to you, so let's do whatever makes your life easier. Let me know.

works for me. Taking it via your tree is the right way to reach master branch.

Thanks,
Michal

#13 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Quoted 14 lines · Michal Simek
Quoted 12 lines · Simon Glass

Hi MIchal,

Quoted 9 lines · Michal Simek

Hi Simon,

Quoted 5 lines · Simon Glass

With a heavy heart:

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

Are you going to take this patch to your DM tree?

I'm hoping to bring in the Kconfig patches and send a pull request this week, so could do that if it suits. But it is up to you, so let's do whatever makes your life easier. Let me know.

works for me. Taking it via your tree is the right way to reach master branch.

Applied to u-boot-dm, thanks!

#14 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

In reply to Michal Simek

Hi Michal,

Quoted 39 lines · On Tue, 3 Feb 2015 10:11:39 +0100 Michal Simek <[email protected]> wrote:

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 1 line
Quoted 4 lines
Quoted 1 line

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

It is not what I meant.

If SPL can directly load the main u-boot image
to the DRAM address where it is linked,
we do not relocate the code in the main image.

Best Regards
Masahiro Yamada

#15 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Quoted 50 lines · Masahiro Yamada

Hi Michal,

Quoted 39 lines · On Tue, 3 Feb 2015 10:11:39 +0100 Michal Simek <[email protected]> wrote:

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 1 line
Quoted 4 lines
Quoted 1 line

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

Quoted 6 lines

It is not what I meant.

If SPL can directly load the main u-boot image
to the DRAM address where it is linked,
we do not relocate the code in the main image.

Current behavior is that SPL is reading u-boot.img entry point which
can be in any location and jump to it and u-boot self relocate to the end of memory.
If SPL adds u-boot directly to the location where it should run after relocation then relocation is not needed.
To ensure this capability (based on my poor GOT/REL/RELA) experience it means that SPL loads u-boot to that location and patch REL/RELA section based on this location and internal relocation should be skipped.
This is definitely doable for REL/RELA case and it can also speedup boot process (I don't think there is easy way how to solve this with just GOT relocation because of that MANUAL_RELOC code which is patching arrays with function pointers).

Thanks,
Michal

#16 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hello Michal,

Quoted 58 lines · Michal Simek
Quoted 50 lines · Masahiro Yamada

Hi Michal,

Quoted 39 lines · On Tue, 3 Feb 2015 10:11:39 +0100 Michal Simek <[email protected]> wrote:

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 1 line
Quoted 4 lines
Quoted 1 line

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

Quoted 6 lines

It is not what I meant.

If SPL can directly load the main u-boot image
to the DRAM address where it is linked,
we do not relocate the code in the main image.

Quoted 8 lines

Current behavior is that SPL is reading u-boot.img entry point which
can be in any location and jump to it and u-boot self relocate to the end of memory.
If SPL adds u-boot directly to the location where it should run after relocation then relocation is not needed.
To ensure this capability (based on my poor GOT/REL/RELA) experience it means that SPL loads u-boot to that location and patch REL/RELA section based on this location and internal relocation should be skipped.

IOW, that SPL perform the work of relocate_code() in U-Boot -- at least, on ARM, where REL/RELA is used.

Quoted 1 line

This is definitely doable for REL/RELA case and it can also speedup boot process

Not sure about the speed-up, but never mind.

Quoted 2 lines

(I don't think there is easy way how to solve this with just GOT relocation because of that MANUAL_RELOC code which is patching arrays with function pointers).

Even without importing SPL in the equation, switching from GOT to REL/RELA has enourmous advantages.

Quoted 2 lines

Thanks,
Michal

Amicalement,

Signature · 1 line
Albert.
#17 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

Hi Albert,

Quoted 71 lines · Albert ARIBAUD

Hello Michal,

Quoted 58 lines · Michal Simek
Quoted 50 lines · Masahiro Yamada

Hi Michal,

Quoted 39 lines · On Tue, 3 Feb 2015 10:11:39 +0100 Michal Simek <[email protected]> wrote:

Hi Simon,

Quoted 21 lines · Masahiro Yamada

Hi.

Quoted 15 lines · On Mon, 2 Feb 2015 16:57:15 -0700 Simon Glass <[email protected]> wrote:

Hi Michal,

Quoted 11 lines · Michal Simek

Targets with CONFIG_NEEDS_MANUAL_RELOC do not use REL/RELA relocation (mostly only GOT) where functions aray are not updated. This patch is fixing function pointers for DM core and serial-uclass to ensure that relocated functions are called.

---

 drivers/core/root.c            | 64 ++++++++++++++++++++++++++++++++++++++++++
 drivers/serial/serial-uclass.c | 16 +++++++++++
 2 files changed, 80 insertions(+)
Quoted 7 lines
Quoted 2 lines

How long will we have to carry this patch? It seems that if we add any new driver we will have to add more code like this?

This patch is unfortunate.
Can we discontinue CONFIG_NEEDS_MANUAL_RELOC some day?

This patch (or similar one) has to be alive when we have platform which requires CONFIG_NEEDS_MANUAL_RELOC for full u-boot. There is an option to move to REL/RELA but the question is if all platforms have it/support it. Unfortunately I think that it will be in the tree for a long time.

Quoted 1 line
Quoted 4 lines
Quoted 1 line

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

Quoted 6 lines

It is not what I meant.

If SPL can directly load the main u-boot image
to the DRAM address where it is linked,
we do not relocate the code in the main image.

Quoted 8 lines

Current behavior is that SPL is reading u-boot.img entry point which
can be in any location and jump to it and u-boot self relocate to the end of memory.
If SPL adds u-boot directly to the location where it should run after relocation then relocation is not needed.
To ensure this capability (based on my poor GOT/REL/RELA) experience it means that SPL loads u-boot to that location and patch REL/RELA section based on this location and internal relocation should be skipped.

Quoted 2 lines

IOW, that SPL perform the work of relocate_code() in U-Boot -- at least, on ARM, where REL/RELA is used.

Maybe we don't understand each other. SPL is not relocate self to any other location. It just copies full u-boot to memory and jump to it and then full U-Boot relocate self to the end of memory.

Quoted 4 lines
Quoted 2 lines

This is definitely doable for REL/RELA case and it can also speedup boot process

Not sure about the speed-up, but never mind.

What I wanted to describe is that imagine case that SPL will does a little bit more steps when full u-boot should be loaded. It loads full u-boot to the end of memory and fix rel/rela section before passing control to it.
IMHO (and not expert) removes one u-boot copy which current full u-boot does. (On the other hand if you want speedup boot you can boot OS directly).

Quoted 6 lines
Quoted 3 lines

(I don't think there is easy way how to solve this with just GOT relocation because of that MANUAL_RELOC code which is patching arrays with function pointers).

Even without importing SPL in the equation, switching from GOT to REL/RELA has enourmous advantages.

I believe you - I haven't finish this on Microblaze as we talked on IRC some time ago. I have to do what arm64 is doing in tools/relocate-rela.c.

Thanks,
Michal

#18 lore raw

RFC PATCH 1/1 dm: Add support for all targets which requires MANUAL_RELOC

In reply to Albert ARIBAUD

Hi Albert,

Quoted 13 lines
Quoted 13 lines
Quoted 13 lines
Quoted 6 lines
Quoted 3 lines

If we use SPL, we do not have to relocate code, I think.

SPL doesn't have relocation that's why this code is not used there.

It is not what I meant.

If SPL can directly load the main u-boot image
to the DRAM address where it is linked,
we do not relocate the code in the main image.

Quoted 8 lines
Quoted 8 lines

Current behavior is that SPL is reading u-boot.img entry point which
can be in any location and jump to it and u-boot self relocate to the end of memory.
If SPL adds u-boot directly to the location where it should run after relocation then relocation is not needed.
To ensure this capability (based on my poor GOT/REL/RELA) experience it means that SPL loads u-boot to that location and patch REL/RELA section based on this location and internal relocation should be skipped.

I've always thought the logical approach would be for SPL to calculate where U-Boot would end up in RAM, copy it from wherever (NAND, NOR, MMC, whatever) to the final location, perform the relocation fixups, then jump straight into U-Boot. It's pretty much what the Dynamic Loader does (particularly with address space layout randomization)

Quoted 7 lines

IOW, that SPL perform the work of relocate_code() in U-Boot -- at least, on ARM, where REL/RELA is used.

Quoted 1 line

This is definitely doable for REL/RELA case and it can also speedup boot process

Not sure about the speed-up, but never mind.

It will save a RAM to RAM copy of u-boot's text and data sections. However, running the relocation code in SPL may be even slower (depending on configuration of caches).

Quoted 5 lines
Quoted 2 lines

(I don't think there is easy way how to solve this with just GOT relocation because of that MANUAL_RELOC code which is patching arrays with function pointers).

Even without importing SPL in the equation, switching from GOT to REL/RELA has enourmous advantages.

I must admit, I'm really rusty with GOT vs REL/RELA - anyone care to give me a refresher?

But yes, only having to deal with REL/RELA sections will make the relocation procedures much simpler

Regards,

Graeme