From: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
To: Tom Rini <trini@konsulko.com>,
Joe Hershberger <joe.hershberger@ni.com>,
Ramon Fried <rfried.dev@gmail.com>,
MD Danish Anwar <danishanwar@ti.com>,
Jai Luthra <j-luthra@ti.com>,
Siddharth Vadapalli <s-vadapalli@ti.com>,
Nishanth Menon <nm@ti.com>, Udit Kumar <u-kumar1@ti.com>,
Suman Anna <s-anna@ti.com>
Cc: u-boot@lists.denx.de, u-boot@ew.tq-group.com,
Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Subject: [PATCH 1/5] net: eth-uclass: guard against reentrant eth_init()/eth_halt() calls
Date: Fri, 26 Apr 2024 10:02:24 +0200 [thread overview]
Message-ID: <de7fe2da7ae4a85252d63904fae61585e4641936.1714117337.git.matthias.schiffer@ew.tq-group.com> (raw)
With netconsole, any log message can result in an eth_init(), possibly
causing an reentrant call into eth_init() if a driver's ops print
anything:
eth_init() -> driver.start() -> printf() -> netconsole -> eth_init()
eth_halt() -> driver.stop() -> printf() -> netconsole -> eth_init()
Rather than expecting every single Ethernet driver to handle this case,
prevent the reentrant calls in eth_init() and eth_halt().
The issue was noticed on an AM62x board, where a bootm after
simultaneous netconsole and TFTP would result in a crash.
Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
---
net/eth-uclass.c | 40 ++++++++++++++++++++++++++++++++--------
1 file changed, 32 insertions(+), 8 deletions(-)
diff --git a/net/eth-uclass.c b/net/eth-uclass.c
index 3d0ec91dfa4..193218a328b 100644
--- a/net/eth-uclass.c
+++ b/net/eth-uclass.c
@@ -48,6 +48,8 @@ struct eth_uclass_priv {
/* eth_errno - This stores the most recent failure code from DM functions */
static int eth_errno;
+/* Are we currently in eth_init() or eth_halt()? */
+static bool in_init_halt;
/* board-specific Ethernet Interface initializations. */
__weak int board_interface_eth_init(struct udevice *dev,
@@ -285,11 +287,19 @@ U_BOOT_ENV_CALLBACK(ethaddr, on_ethaddr);
int eth_init(void)
{
- char *ethact = env_get("ethact");
- char *ethrotate = env_get("ethrotate");
struct udevice *current = NULL;
struct udevice *old_current;
int ret = -ENODEV;
+ char *ethrotate;
+ char *ethact;
+
+ if (in_init_halt)
+ return -EBUSY;
+
+ in_init_halt = true;
+
+ ethact = env_get("ethact");
+ ethrotate = env_get("ethrotate");
/*
* When 'ethrotate' variable is set to 'no' and 'ethact' variable
@@ -298,8 +308,10 @@ int eth_init(void)
if ((ethrotate != NULL) && (strcmp(ethrotate, "no") == 0)) {
if (ethact) {
current = eth_get_dev_by_name(ethact);
- if (!current)
- return -EINVAL;
+ if (!current) {
+ ret = -EINVAL;
+ goto end;
+ }
}
}
@@ -307,7 +319,8 @@ int eth_init(void)
current = eth_get_dev();
if (!current) {
log_err("No ethernet found.\n");
- return -ENODEV;
+ ret = -ENODEV;
+ goto end;
}
}
@@ -324,7 +337,8 @@ int eth_init(void)
priv->state = ETH_STATE_ACTIVE;
priv->running = true;
- return 0;
+ ret = 0;
+ goto end;
}
} else {
ret = eth_errno;
@@ -344,6 +358,8 @@ int eth_init(void)
current = eth_get_dev();
} while (old_current != current);
+end:
+ in_init_halt = false;
return ret;
}
@@ -352,17 +368,25 @@ void eth_halt(void)
struct udevice *current;
struct eth_device_priv *priv;
+ if (in_init_halt)
+ return;
+
+ in_init_halt = true;
+
current = eth_get_dev();
if (!current)
- return;
+ goto end;
priv = dev_get_uclass_priv(current);
if (!priv || !priv->running)
- return;
+ goto end;
eth_get_ops(current)->stop(current);
priv->state = ETH_STATE_PASSIVE;
priv->running = false;
+
+end:
+ in_init_halt = false;
}
int eth_is_active(struct udevice *dev)
--
TQ-Systems GmbH | Mühlstraße 2, Gut Delling | 82229 Seefeld, Germany
Amtsgericht München, HRB 105018
Geschäftsführer: Detlef Schneider, Rüdiger Stahl, Stefan Schneider
https://www.tq-group.com/
next reply other threads:[~2024-04-26 8:02 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-04-26 8:02 Matthias Schiffer [this message]
2024-04-26 8:02 ` [PATCH 2/5] net: ti: am65-cpsw-nuss: avoid errors due to imbalanced start()/stop() Matthias Schiffer
2024-04-26 8:02 ` [PATCH 3/5] net: ti: am65-cpsw-nuss: fix error handling for "RX dma add buf failed" Matthias Schiffer
2024-04-26 8:02 ` [PATCH 4/5] dma: ti: k3-udma: add missing <net.h> include Matthias Schiffer
2024-04-26 8:02 ` [PATCH 5/5] dma: ti: k3-udma: invalidate prepared buffers before pushing to recv ring Matthias Schiffer
2024-05-03 18:42 ` Sverdlin, Alexander
2024-04-26 8:09 ` [PATCH 1/5] net: eth-uclass: guard against reentrant eth_init()/eth_halt() calls Siddharth Vadapalli
2024-05-15 23:28 ` Tom Rini
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=de7fe2da7ae4a85252d63904fae61585e4641936.1714117337.git.matthias.schiffer@ew.tq-group.com \
--to=matthias.schiffer@ew.tq-group.com \
--cc=danishanwar@ti.com \
--cc=j-luthra@ti.com \
--cc=joe.hershberger@ni.com \
--cc=nm@ti.com \
--cc=rfried.dev@gmail.com \
--cc=s-anna@ti.com \
--cc=s-vadapalli@ti.com \
--cc=trini@konsulko.com \
--cc=u-boot@ew.tq-group.com \
--cc=u-boot@lists.denx.de \
--cc=u-kumar1@ti.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).